David Weisburd
Ernesto U. Savona
Badi Hasisi
Francesco Calderoni Editors
Understanding
Recruitment
to Organized
Crime and
Terrorism
Understanding Recruitment to Organized Crime
and Terrorism
David Weisburd • Ernesto U. Savona
Badi Hasisi • Francesco Calderoni
Editors
Understanding Recruitment
to Organized Crime and
Terrorism
Editors
David Weisburd
Institute of Criminology
Faculty of Law
Hebrew University of Jerusalem
Jerusalem, Israel
Department of Criminology
Law and Society
George Mason University
Fairfax, Virginia, USA
Ernesto U. Savona
Transcrime
Università Cattolica del Sacro Cuore
Milan, Italy
Francesco Calderoni
Transcrime
Università Cattolica del Sacro Cuore
Milan, Italy
Badi Hasisi
Institute of Criminology
Faculty of Law
Hebrew University of Jerusalem
Jerusalem, Israel
ISBN 978-3-030-36638-4 ISBN 978-3-030-36639-1 (eBook)
https://doi.org/10.1007/978-3-030-36639-1
© Springer Nature Switzerland AG 2020
Chapter “Delinquent development, employment and income in a sample of Dutch organized crime
offenders: Shape, content, and correlates of delinquent trajectories from age 12 to 65” is licensed under
the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/
licenses/by/4.0/). For further details see license information in the chapter.
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the
editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Contents
Introduction������������������������������������������������������������������������������������������������������ 1
David Weisburd, Ernesto U. Savona, Badi Hasisi,
and Francesco Calderoni
Part I Factors Affecting Radicalization and Recruitment to Terrorism
What Is the State of the Quantitative Literature on Risk Factors
for Radicalization and Recruitment to Terrorism?�������������������������������������� 25
Michael Wolfowicz, Yael Litmanovitz, David Weisburd, and Badi Hasisi
Resilience against Political and Religious Extremism,
Radicalization, and Related Violence: A Systematic Review
of Studies on Protective Factors���������������������������������������������������������������������� 55
Friedrich Lösel, Doris Bender, Irina Jugl, and Sonja King
Terrorist Recidivism in Israel: Rates, Patterns and Risk Factors �������������� 85
Tomer Carmel, Michael Wolfowicz, Badi Hasisi, and David Weisburd
Unintended Negative Outcomes of Counter-Terrorism Policing:
Procedural (in)Justice and Perceived Risk of Recruitment
into Terrorism�������������������������������������������������������������������������������������������������� 105
Justice Tankebe
The Link Between Prior Criminal Record and Violent Political
Extremism in the United States���������������������������������������������������������������������� 121
Michael A. Jensen, Aaron Safer-Lichtenstein, Patrick A. James,
and Gary LaFree
v
vi
Contents
Testing a Threat Model of Terrorism: A Multi-method Study About
Socio-­Economic and Psychological Influences on Terrorism
Involvement in the Netherlands���������������������������������������������������������������������� 147
Vanja Ljujic, Inge Versteegt, Frank Weerman, Fabienne Thijs,
Jan-­Willem van Prooijen, Fatima el Bouk, and Steve van de Weijer
Part II Factors Affecting Recruitment to Organized Crime
Systematic Review of the Social, Psychological and Economic Factors
Relating to Involvement and Recruitment into Organized Crime�������������� 175
Tommaso Comunale, Francesco Calderoni, Martina Marchesi,
Elisa Superchi, and Gian Maria Campedelli
Socio-Economic Inequalities and Organized Crime:
An Empirical Analysis ������������������������������������������������������������������������������������ 205
Michele Battisti, Giovanni Bernardo, Antri Konstantinidi,
Andros Kourtellos, and Andrea Mario Lavezzi
The Criminal Careers of Italian Mafia Members���������������������������������������� 241
Ernesto U. Savona, Francesco Calderoni, Gian Maria Campedelli,
Tommaso Comunale, Marco Ferrarini, and Cecilia Meneghini
Investigating the Psychological Profile of Organized Crime Members ������� 269
Gerardo Salvato, Maria Laura Fiorina, Daniela Ovadia,
Gabriele De Maio, Elisa Francescon, and Gabriella Bottini
Traffickers in Transit: Analysing the Logistics and Involvement
Mechanisms of Organised Crime at Logistical Nodes in the
Netherlands: Empirical Results of the Dutch Organised
Crime Monitor ������������������������������������������������������������������������������������������������ 277
Renushka Madarie and Edwin W. Kruisbergen
Delinquent Development, Employment and Income in a Sample
of Dutch Organized Crime Offenders: Shape, Content,
and Correlates of Delinquent Trajectories from Age 12 to 65 �������������������� 309
Victor van der Geest, M. Vere van Koppen, and Edward R. Kleemans
Index������������������������������������������������������������������������������������������������������������������ 337
About the Editors
David Weisburd is a Distinguished Professor at George Mason University and
Director of the Center for Evidence-Based Crime Policy. He is also the Walter
E. Meyer Professor of Law and Criminal Justice at the Hebrew University Faculty
of Law in Jerusalem. He serves as Chief Science Advisor at the Police Foundation
in Washington DC and is Chair of its Research Advisory Committee. He has
received many awards for his contributions to criminology and crime prevention.
These include the Stockholm Prize in Criminology (2010); the Klachky Family
Prize for the Advancement of the Frontiers of Science (2011); the Jerry Lee Award
for Lifetime Achievement in Experimental Criminology (Division of Experimental
Criminology, ASC, 2014); the Robert Boruch Award for distinctive contributions to
research that influences public policy of the Campbell Collaboration (2014); the
Sutherland Award for “outstanding contributions to the discipline of criminology”
(ASC, 2014); the Israel Prize (generally regarded as the State of Israel’s highest
civilian honor) (2015); the Mentoring Award for “excellence in mentorship in the
discipline of Criminology and Criminal Justice” (ASC, 2016); and the August
Vollmer Award for contributions to the prevention of crime (ASC, 2017).
Ernesto U. Savona is Director of TRANSCRIME, Joint Research Centre of the
Università Cattolica del Sacro Cuore and the Università degli Studi di Trento, and
since 2003, he is Professor of Criminology at the Università Cattolica del Sacro
Cuore in Milan. From 1986 to 2002, he was Professor of Criminology at the Faculty
of Law at the Università degli Studi di Trento. In 2003, he was appointed Editor in
Chief of the European Journal on Criminal Policy and Research, an international
refereed Journal published by Springer. He is also Editor in Chief of the series
Research in Brief in International and Comparative Criminology published by
Springer. He has been consultant to the United Nations, Council of Europe, the
European Union, and various national governments. He was also nominated
President of the European Society of Criminology for the years 2003/2004. In June
2011, he was appointed Chair for the term 2011–2012 of the Global Agenda Council
on Organized Crime del World Economic Forum.
vii
viii
About the Editors
Badi Hasisi serves as an Associate Professor and Chair of the Institute of
Criminology, Faculty of Law, The Hebrew University. His work focuses on the
interaction between community and the criminal justice agencies, with specific
emphasis on the particular problems faced by minority groups with the criminal
justice system. He also specializes in homeland security ad terrorism. In the last
3 years, he received about 900,000 € from the European Commission to study terrorism. He also specializes in evaluating innovations in the criminal justice system.
Prof. Hasisi also specializes in the legal history of Mandate Palestine. His recent
study focuses on analyzing the reaction of the British court system during the
Mandate in Palestine to family honor crimes, especially among the Arab community. He received the 2018 best article prize of the Israeli Organization of Law and
History and the Fattal Prize for Excellence in Legal Research and Criminology.
Prof. Hasisi also serves as the Executive Editor of the Journal of Quantitative
Criminology.
Francesco Calderoni is Associate Professor at the Faculty of Social and Political
Sciences at Università Cattolica del Sacro Cuore and Researcher at Transcrime Joint
Research Centre of the Università Cattolica del Sacro Cuore, the University of
Bologna, and the University of Perugia. He is Coordinator of the International
Ph.D. in Criminology and of the master curriculum in Policies for Security at the
M.Sc. in Public Policies at the Università Cattolica in Milan. He is Associate Editor
at Global Crime. He was awarded the Ambrosoli Prize 2010 by the Milan City
Council for the best Ph.D. thesis and the Canada-Italy innovation award 2014 by the
Canadian Embassy in Italy for the application of network analysis methods to criminal organizations.
Introduction
David Weisburd, Ernesto U. Savona, Badi Hasisi, and Francesco Calderoni
The New York Mafia has launched an unprecedented recruitment drive after deaths, defections and the unwelcome interventions of the FBI have cut the number of “wise guys” under
its command by more than 10 per cent in a year. Numbers on the streets fell from 634 to 570
during 2000, according to the FBI, prompting the Italian crime families that still run many
of the city’s drugs, gambling, extortion and prostitution rackets to “open the books” to new
members… The existence of the recruitment drive was confirmed to the authorities by an
informant, Michael “Cookie” Durso.”
Short-staffed Mafia makes a job offer you can’t refuse. The Daily Telegraph
(Leith 2002)
Britain’s most senior counter-terrorism officer has said the police and security services are
no longer enough to win the fight against violent extremism, and the UK must instead
improve community cohesion, social mobility and education. In his first major interview
since taking up his post last year, the Metropolitan police assistant commissioner Neil Basu
told the Guardian that up to 80% of those who wanted to attack the UK were British-born
or raised, which strongly indicated domestic social issues were among the root causes.
Grievances held by people who were “malleable” to terrorist recruitment were highly dangerous, he said, calling for sociologists and criminologists to take a leading role in helping
police tackle the problem.
Counter-terror chief says policing alone cannot beat extremism. The Guardian
(Dodd 2019)
Organized crime and terrorism stand as two of the key threats to social organization and government in Western countries today (Shaw and Mahadevan 2018).
Organized crime stands behind many illegal enterprises that cater to public desires
that the government seeks to control. These include prostitution, drugs, gambling,
and protection rackets, which have formed a key aspect of organized criminal
D. Weisburd (*)
Institute of Criminology, Faculty of Law, Hebrew University of Jerusalem, Jerusalem, Israel
Department of Criminology, Law and Society, George Mason University,
Fairfax, Virginia, USA
e-mail: dweisbur@gmu.edu
E. U. Savona · F. Calderoni
Transcrime, Università Cattolica del Sacro Cuore Transcrime, Milan, Italy
B. Hasisi
Institute of Criminology, Faculty of Law, Hebrew University of Jerusalem, Jerusalem, Israel
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_1
1
2
D. Weisburd et al.
behaviour for the last century or more (Shelley 1995; Albanese 2014). But organized crime has also been key in creating competitive economies for goods that are
highly taxed, like cigarette sales or liquor (Savona et al. 2017). And importantly,
organized crime groups have often resorted to violence, infiltrated the legal ­economy,
and undermined legal and political institutions to protect their control over these
enterprises (Fleenor 2003; Savona et al. 2016).
Terrorism has served as a more recent threat in Western democracies, though terrorism has operated in one form or another for centuries or even millennia (Rapoport
1983, 2004). Already in ancient times Cicero rallied the Roman government to act
harshly against terrorism, when Catiline conspired to assassinate government officials and set the city ablaze.1 Nevertheless, it is only in more recent times, especially
since the events of 9/11, that terrorism has become a more central focus of policy in
Western countries.
The goals of terrorists differ from organized criminals, in that their primary interests are in changing political or social systems, rather than simply protecting their
economic or social control over specific industries. Nonetheless, terrorism, whether
left or right-wing or religiously motivated, and organized criminal organizations,
whether mafias, gangs, or other entities, have impacted social order in most Western
democracies--whether reflected in heightened levels of fear among citizens who
become afraid to carry out their daily routines (e.g. Mesch 2000; Malik et al. 2018),
or changes in security procedures that restrict everyday access to public and private
institutions (White 1998) or increases in aggressive security procedures against specific ethnic, national, religious or racial groups (Wolfendale 2007). Additionally,
different sectors of the economy can potentially suffer damages as the result of both
terrorism (e.g. Berrebi and Klor 2010; Aslam et al. 2018) and organized crime (e.g.
Lavezzi 2014; Calderoni 2014; Pinotti 2015). The impacts of terrorism, like organized crime, are broadly spread across the social systems of modern democracies.
In recent years, multinational governmental bodies have placed an increasing
focus on the need to cooperate to combat both organized crime and terrorism,
reflected in multiple UN resolutions. A host of organizations have either been created for this expressed purpose, or have adjusted their scope and focus towards such
objectives. For example, INTERPOL and EUROPOL expanded their mandate to
include deeper cooperation on combatting both organized crime and terrorism following the events of 9/11 (Deflem 2007). In recent years, organizations such as the
United Nations Office on Drugs and Crime (UNODC) have taken on an increased
interest not only in transnational organized crime networks but in issues pertaining
to terrorism as well. Meanwhile, a host of other organizations, such as the Action
against Terrorism Unit (ATU) of the Organization for Security and Co-operation in
Europe (OSCE), have been created for the express purpose of combatting organized
crime and terrorism, and specifically facilitating cooperation towards these goals).
M. Tullius Cicero. The Orations of Marcus Tullius Cicero, literally translated by C. D. Yonge,
B. A. London. Henry G. Bohn, York Street, Covent Garden. 1856.
1
Introduction
3
Both types of crime cross traditional jurisdictional boundaries. Many of the
activities of organized crime now cross national borders, whether they involve
human trafficking, or the networks of distribution of drugs or guns, or money laundering and other financial crimes (Kleemans, Soudijn and Weenink 2013). Similarly,
terrorists no longer focus simply on the immediate, local or national grievances that
have often sparked their emergence. Not only do terrorists’ grievances, doctrines,
and movements cross national borders, but even terrorist actors themselves. The
skyjackings of the 1960’s demonstrated to the western world that terrorists were
prepared to carry out attacks on ‘foreign soil’, and that terrorism had become ‘internationalized’ (Hasisi et al. 2019). As Attorney General John Ashcroft commented in
2002, “The menace of terrorism knows no borders, political or geographic.”
A key question is whether there is reason to study organized crime and terrorism
in a single research program. The fact that they both raise key threats to social order,
does not mean that they have enough commonality to encourage academic focus, or
similar prevention approaches. While for several decades the study of organized
crime and terrorism have proceeded in parallel, since the early 1990’s there has been
an emerging interest in research on the overlaps between these two phenomena.
Some of this research emphasized the important links between organized crime and
terrorism, not only conceptually, but also in terms of the actors, and at the levels of
activities, cooperation and organization (Bassiouni 1990; Schmid 1996; Taylor and
Horgan 1999; Williams 1998; Clarke and Newman 2006; Decker and Pyrooz 2011,
2014; Basra and Neumann 2017). Additionally, some scholars argued that organized crime groups and terrorist groups may interchangeably engage in both criminal and terroristic activities, and that both types of groups may even cooperate with
each other in furtherance of their mutual and exclusive goals (Makarenko 2004;
Mullins 2009; Schmid 2018).
Within this growing research interest, however, some scholars have also cautioned against overstating the overlap between terrorism and organized crime
(Hutchinson and O’malley 2007; Dishman, 2001, 2005), emphasizing that violent
extremists and gang members in the US for example differ in terms of several personal characteristics (Pyrooz et al. 2018), and that there remain important differences in their origins and goals (Desmond and Hussain 2017; Felbab-Brown 2019).
Overall, while the possible overlap between terrorism and organized crime is certainly a central issue in recent study of complex forms of group criminality, the
debate on the theoretical, conceptual, and actual extent of overlap is still open and
will likely be settled through “painstaking empirical research” (Bovenkerk and
Chakra 2005, 13). With the purpose of contributing to this debate, our research program sought to contribute knowledge both to our understanding of recruitment into
organized crime and terrorism.
Recruitment has long been a key issue for control and study of gangs, although
less so for organized crime and terrorism (Densley 2012; Smith 2014; Finelli 2019;
Leukfeldt and Kleemans 2019). Yet, as the news articles above suggest, whether in
the case of organized crime or terrorism, recruitment raises important concerns for
social control of these phenomena. In New York, during a period in the early 2000s,
when organized crime groups were under tremendous pressure from police and
4
D. Weisburd et al.
prosecutors, the Mafia sought to increase recruitment. Recently, police agencies
have begun to recognize that recruitment to terrorism is intrinsically linked to the
social grievances and social conditions of communities in Western democracies.
Recruitment is a key issue in the long term control of terrorism and organized crime.
And at least from initial studies it appears that for both organized crime and terrorism, potential recruits tend to be drawn from the same general pool of the p­ opulation
(Basra and Neumann 2017), and that their pathways to recruitment have strong
similarities (Benavides et al. 2016; Vishnevetsky 2009).
But the body of research on recruitment remains relatively sparse. Over the years
it has been estimated that as little as a few percent of radicalization research
(Neumann and Kleinmann 2013), and terrorism research more broadly (Sageman
2014) is quantitative or data driven. Importantly, research on the economic, social,
and psychological risk and protective factors for recruitment to terrorism have
largely been limited to a small set of socio-demographic factors (Stern 2016). With
respect to organized crime, the majority of the research has been qualitative, with
few quantitative studies. Even among the quantitative literature, the focus has primarily been on the prevalence, or costs of organized crime, and little attention has
been paid to the offenders themselves, let alone the individual-level risk factors for
recruitment to organized crime (Thompson 2019). While research has highlighted
that there are important similarities between organized crime and terrorism offenders, and with ordinary criminals, recent findings suggest that there may also be
important differences in the key risk factors that predict recruitment for both organized crime offenders and terrorism offenders (Blokland et al. 2019; Liem et al.
2018; Pyrooz et al. 2018). The lack of rigorous studies into recruitment to organized
crime (Levi and Maguire 2004) and terrorism (Lum, Kennedy and Shirley 2006)
means that policies are often not evidence-based, or may be based on theoretical
assumptions, or assumptions drawn from a relatively small number of case studies.
What are the factors that lead to recruitment to organized crime and terrorism,
and how are they similar or different? What role do situational factors play in
recruitment, as contrasted with social or psychological causes? What impacts do
responses of society to these phenomena have upon populations that are considered
high risk by social control agents? These were the key questions that we sought to
answer in the context of a large European Horizon 2020 research and innovation
funding program. In order to accomplish this, the research consortium included a
wide spectrum of partners from the EU, US, Switzerland, and Israel, as well as
representatives from key institutions, such as the UNODC and EUROPOL.
The “PROTON” (Modelling the Processes Leading to Organized Crime and
Terrorism) consortium was led by the TRANSCRIME Centre at Università Cattolica
del Sacro Cuore in Milan, Italy, and co-coordinated by the Institute of Criminology
at the Hebrew University of Jerusalem. In order to capitalize on the strengths of
each institute, TRANSCRIME led the organized crime track of the project, and
HUJI led the terrorism track. Each track consisted of a series of independent research
projects. Project PROTON placed a focus on conducting unique, original studies,
employing new and original datasets. Primarily, the studies were focussed on identifying the micro, meso, and macro level factors that increase the risk of recruitment
Introduction
5
to organized crime, and radicalization and recruitment to terrorism. From its conception, the individual research tasks, as well as the project overall, were designed
specifically to fill key gaps that were identified in the existing body of knowledge.
Our goal in this volume is to present key findings from the project that illuminate
our understanding of recruitment to organized crime and terrorism. As detailed
below, the papers included add substantive and significant knowledge to our understanding of these questions focusing on the economic, social, and psychological
factors that increase the likelihood of recruitment. Below, we first briefly review the
existing literature on recruitment to terrorism and organized crime, identifying key
gaps in the literature. We then turn to the specific contributions in the volume and
summarize how they have added to the knowledge base in this area. Finally, we
revisit the question of similarities and differences between factors influencing
recruitment based on the papers in the volume, and try to draw broader conclusions
about how this knowledge can inform societal responses to the problem of
recruitment.
Radicalization and Recruitment to Terrorism
Radicalization and recruitment to terrorism are two separate but interrelated outcomes. Radicalization generally refers to the development of attitudes, beliefs, and
ideas that justify the use of radical violence to promote change, and thereby which
could lead to radical violence. On the other hand, recruitment is when an individual
has been solicited, or convinced, to carry out illegal activities that fall under the
category or legal definition of terrorism.2 As such, radicalization generally underpins recruitment, or is at least a near universal precursor to it. Nevertheless, only a
very small percentage of radicalized individuals will ever turn to terrorism
(McCauley and Moskalenko 2017; Neumann 2013). A key goal of radicalization
research is therefore to identify what differentiates non-violent from violent radicals
(LaFree and Freilich 2016).
Many scholars now believe that there must be certain risk factors that explain
why some members of an aggrieved group are led to justify radical violence, while
the majority do not. It also must be that certain risk factors could explain why a
small proportion of radicals go on to engage in terrorism offending, whilst most do
not (Armstrong et al. 2019; Haddad 2017; Sarma 2017; Stern 2016). The best methods for identifying which factors may be the most important for explaining these
different outcomes is to compare radicals with non-radicals, and violent radicals
with non-violent radicals (Knight, Woodward and Lancaster 2017; LaFree and
Frelich 2016). Unfortunately, to date, few studies have taken this approach. The
These definitions are according to those of the European Commission’s strategy for combating
radicalisation and recruitment (2005, 2014). While there is an extensive literature surrounding
the debate over these definitions, the EU definitions are sufficiently broad so as to encapsulate the
primary elements of the phenomena being discussed.
2
6
D. Weisburd et al.
relatively young literature on radicalization is still made up of primarily theoretical,
or qualitative study, with quantitative study representing only a small proportion of
the body of knowledge (Neumann and Kleinmann 2013). While the literature has
identified a range of factors that appear to be correlated with radicalization o­ utcomes,
the evidence remains quite mixed, and certain factors have been given more attention than others (Allan et al. 2015; Bondokji, Wilkinson and Aghabi 2017).
For example, one of the most commonly discussed factors is socio-economic
status. Whilst many believed, and still believe, that poor economic conditions
increase the risk of radicalization and recruitment to terrorism, most studies have
found non-significant or at best weak relationships between individual socio-­
economic status and radicalization outcomes. Similarly, with regards to education,
the evidence is quite mixed. While some studies have identified lower education as
a risk factor, others point to a high prevalence of higher education among recruited
terrorism offenders, and the university campus as a hot-spot of recruitment (Allan
et al. 2015; Bondokji, Wilkinson and Aghabi 2017; Victoroff 2005). Other related
factors, such as unemployment, underemployment, recent job loss or employment
instability, do appear to have more consistent effects. Whilst objective socio-economic status may not be as important as some believe, subjective forms of deprivation, such as collective and individual relative deprivation, which includes elements
of perceived group-based discrimination, have been found to be strong predictors of
radicalization.
The findings concerning subjective beliefs and perceptions are perhaps even
more interesting if contextualized as socio-psychological factors. The study of psychological factors has undergone a pendulum swing since radicalization became a
topic of interest to researchers. While early terrorism scholars often assumed that
terrorists must be ‘mad men’, a growing body of evidence found that the only psychological constant among terror recruits was an overwhelming normality
(Kruglanski and Golec 2004; Merari 2004; Horgan 2005; Gill 2012). As this became
the prevailing wisdom, the study of psychological factors became scarce (Victoroff
2005; Victoroff, Adelman and Matthews 2012; Stern 2016). Yet in recent years a
new focus has been placed on the role of mental-illness more broadly, and psychological and personality traits in particular (Gill and Corner 2017). While these factors are well known to criminologists and strong predictors of a range of criminal
attitudes and behaviors, the number of studies that examine these types of factors
with respect to radicalization and recruitment to terrorism is still small.
Recruitment to Organized Crime
The emergence of criminal organizations has been explained as being driven by a
wide range of factors, including social, economic and psychological ones (see e.g.
Lyman and Potter 2006). But scholars have generally argued that social factors
occupy the primary role in guiding the decision to become involved in an organized
crime (OC) network. Indeed, OC has particular characteristics that differentiate it
Introduction
7
from high-volume crime: in OC, social relations are more important; there is a
transnational dimension; the activities are more complex and they require several
co-offenders and specific expertise (Cornish and Clarke 2002; Kleemans and de
Poot 2008). Relations with co-offenders are important, but not always sufficient,
because these offenders may not possess the necessary capabilities. Contacts with
the legal world are also important for activities of this kind. Many people lack these
contacts and expertise, and some acquire them only later in life, e.g. through their
professional activities and contacts.
Kleemans and van Koppen (2014) identified three key aspects guiding the choice
of involvement in OC. Firstly, besides the presence of career criminals in the world
of OC, researchers have pointed to the presence of a group of ‘late onset’ offenders.
This finding is robust across several different criminal activities (drugs, fraud, and
other activities) and different roles in criminal groups (leaders, coordinators, and
lower-level suspects). This presence contradicts the general ideas that crimes in
adolescence precede crimes at a later age, and that individuals’ characteristics and
long-term risk factors are the main explanations of a life in crime. Secondly, various
involvement mechanisms play a role in OC. Besides deliberate recruitment, four
other involvement mechanisms are important: social ties and the ‘social snowball
effect’, work ties, leisure activities and sidelines, and life events (including financial
setbacks). Thirdly, work relations and work settings may be the breeding grounds of
OC activities, particularly for cross-border crimes.
Economic factors have generally received less attention. However, the recent
literature (see Lavezzi 2008; 2014, and the references therein) has identified three
economic factors determining the evolution of OC. Firstly, the structure of the economy can provide a favorable environment for typical OC activities, such as extortion and protection. Lavezzi (2008), who draws insights from Schelling (1971),
argues that Sicily is a prime example of a region plagued by the presence of OC, and
links OC to an economy characterized by a large proportion of small firms, large
sectors of traditional/low-tech economic activities, a large construction sector, and
a large public sector. Secondly, imperfections in the credit markets can create opportunities for usury, especially in the absence of the rule of law. Thirdly, the presence
of illegal and informal markets, where legal institutions cannot be invoked to establish rules, create opportunities for OC to provide an alternative rule of law and also
to act as a law enforcer.
Surprisingly, the economic literature has paid little attention to the role of
inequality. Only one study we know of examined the effects of poverty and deprivation, such as the emergence of inner city gangs (e.g. Lyman and Potter 2006). Yet
inequality can lead to the emergence of OC. Indeed, data from Italian regions show
that in the Southern regions, where criminal organizations are pervasive, income
inequality measured by the Gini index is relatively high. This phenomenon has deep
historical roots, which have been analyzed, for example, in terms of the role of land
ownership in the emergence of OC by Bandiera (2003).
Other factors, such as self-reinforcing mechanisms, may also impact on the
emergence of OC and may affect both inequality and social mobility. Social mobility is typically measured by intergenerational mobility, which quantifies the degree
8
D. Weisburd et al.
of fluidity between the parents’ socio-economic status and the offspring’s socio-­
economic status as adults. Wealthy members of society may reinforce their claims
to resources by utilizing OC and distorting the efficient allocation of physical and
human capital. As a result, the presence of OC may lead to less social mobility and
lower economic growth. Under certain conditions, the presence of OC may generate
poverty traps that prevent the children of low-income families from having any better opportunities than recruitment by the mafia.
As regards psychological factors, members of OC show not only a general predisposition to violence but also more complex cognitive and emotional traits determining their functional roles in OC networks. Nevertheless, few studies have used
neuroscientific tools to identify the principal psychological and cognitive determinants of the complex criminal behaviors that characterize OC. Most studies use as
their tool of investigation only the Psychopathy Check List-Revised (Ostrosky et al.
2011), a controversial instrument with a number of limitations, the most important
being the diagnosis of psychopathy itself (Hare 1999; Edens 2006). Moreover, neuroscientific tools such as neuropsychological and genetic tests and even neuroimaging are widely used in courts to explain (and sometimes justify) criminal behaviors
(Jones et al. 2013). To advance understanding of the psychological determinants of
criminal behaviors in an innovative manner, it is necessary to adopt these new tools
as well, and thus move beyond the tools and models used by classic criminology
(Walsh and Beaver 2009).
The Studies
Our review of present knowledge about recruitment to terrorism and organized
crime suggest the scarcity of studies on key issues in this area of research. The
PROTON studies sought to fill many of the gaps in the literature. The purpose of
this effort in PROTON was to inform the development of simulation models for
testing different types of societal interventions to reduced recruitment. However,
these studies have produced important findings that advance the knowledge base in
this area. And that is the reason for our joining together these contributions in
this volume.
We begin in Part I focusing on the terrorism track of PROTON with systematic
reviews of the existing literature in the chapters “What is the State of the Quantitative
Literature on Risk Factors for Radicalization and Recruitment to Terrorism?” and
“Resilience Against Political and Religious Extremism, Radicalization, and Related
Violence: A Systematic Review of Studies on Protective Factors”. Systematic
reviews of research evidence are characterized by careful searches of the research
literature, including both published and unpublished studies, and researchers in our
case sought to draw general conclusions from work based on meta-analytic statistics (Lipsey and Wilson 2001; Petrosino, Boruch, Farrington, Sherman, and
Weisburd 2003; Weisburd, Farrington, and Gill 2017). In the first review, Wolfowicz,
Litmanovitz, Weisburd and Hasisi identify and describe the makeup of the quantita-
Introduction
9
tive literature on risk factors for radicalization and recruitment to terrorism. The
review identifies that there has been a significant uptick in publications in recent
years, with the majority of the 57 included studies published since 2016. Whilst the
majority of studies focus on Islamic radicalization, a number of studies have
­examined mixed samples, without focusing on any particular form of extremism.
The majority of studies examine radicalization of beliefs, most commonly by
assessing support or justification of different forms of terrorism. Only a small number of studies were identified which compare terrorists with control or comparison
groups of non-terrorists. The findings of the review were that while socio-demographic characteristics are the most commonly explored factors, they had relatively
small effect sizes. On the other hand, traditional criminogenic (e.g. criminal history,
unemployment, low self control, and differential associations) and criminotrophic
factors (e.g. general obedience to the law and strong social bonds) were found to
have the strongest correlations with radicalization and recruitment outcomes.
These findings are consistent with Lösel, Bender, Jugl, and King’s systematic
review on protective factors, which found important overlaps in the types of protective factors against radicalization and recruitment to terrorism with those known
from the literatures on delinquency and crime. The chapter details the evidence collected from 28 studies pertaining to 30 identified protective factors. While the number of studies and factors is somewhat small, the review found significant overlap
with the protective factors known for general resilience and desistance from ordinary crime, including: self-control, adherence to law, acceptance of police legitimacy, illness, non-deviant significant others, positive parenting behavior, good
school achievement, non-violent peers, contact to foreigners, and a basic attachment
to society.
The chapter: “Terrorist Recidivism in Israel: Rates, Patterns and Risk Factors”
looks at recidivism for terrorist offenders. The literature has pointed to this as a key
gap in the body of knowledge for over a decade, and yet little empirical research has
been conducted (Ahmed 2016; Veldhuis and Kessels 2013; Pluchinsky 2008;
Schuurman and Bakker 2016). This is primarily due to a lack of settings in which
sufficient base rates can provide for meaningful analysis. Given that in the coming
years many terrorist offenders in Europe are set to be released, the importance of
such study cannot be underestimated. Taking up this challenge, Carmel, Wolfowicz,
Hasisi and Weisburd examined recidivism to terrorism in Israel, as well as a number
of key factors that affect the likelihood of recidivism. Following the general trends
identified in criminology regarding the rates of recidivism to the same offence compared to overall recidivism, the study identified that the rate of recidivism to terrorism offending is about half of the general recidivism rate in Israel. Examining the
rates more in depth reveals that that recidivism rates were found to increase with
each additional incarceration. Older age at time of release and longer sentences
were both found to be correlated with a lower risk of recidivism.
In recent years there has been an increased interest in the potential for counter-­
terrorism policing policies to have unintended negative outcomes. Some have suggested that hard approaches to counter-terrorism policing, which target or single out
specific minority communities and members, can potentially even lead to a backlash
10
D. Weisburd et al.
effect and increase the risk of radicalization and recruitment to terrorism (Choudhury
and Fenwick 2011; den Boer 2015; Harcourt 2006; LaFree and Ackerman 2009).
Tankebe uses perspectives on procedural justice to provide a framework for understanding this dilemma in the chapter “Unintended Negative Outcomes of CounterTerrorism Policing: Procedural (In)Justice and Perceived Risk of Recruitment into
Terrorism”. He employs an experimental vignette design to examine how the nature
of counter-terrorism practices affect people’s perceptions of the risks of recruitment
into terrorism. A national representative sample of Muslims in the UK was randomly assigned to one of four experimental conditions showing (a) procedurally
just street stops, (b) procedurally unjust street stops, (c) procedurally just counter-­
terrorism raids, or (d) procedurally unjust counter-terrorism raids. After reading the
scenarios participants completed a short questionnaire on the likely effects of the
condition. Procedurally unjust street stops were deemed to cause increased risks of
radicalisation. However, there was no evidence that procedurally unjust counter-­
terrorism raids had similar unintended negative consequences.
One of the largest gaps in the literature relates to the lack of studies that compare
violent and non-violent extremists. Drawing on data from the PIRUS dataset,
Michael Jensen and colleagues examine the role of a number of key risk factors in
differentiating between these two groups in the chapter “The Link Between Prior
Criminal Record and Violent Political Extremism in the United States”. The study
finds that prior criminal history is the single strongest predictor of violent extremist
offending. Moreover, extremist offenders acting upon far right-wing ideologies are
more likely to have criminal histories than other extremists. The study also found
that juvenile criminal offending was a greater risk factor than adult criminal offending for later extremist offending.
In the chapter “Testing a Threat Model of Terrorism: A Multi-method Study
About Socio-Economic and Psychological Influences on Terrorism Involvement in
the Netherlands”, Frank Weerman and colleagues provide a mixed methods study to
investigate the role of socio-economic and psychological factors and processes
leading to involvement in terrorism. The study includes a qualitative component in
which the researchers compare interviews conducted in the Netherlands with four
detained terrorism offenders, and a number of ordinary criminal offenders and other
informants. The study’s quantitative component includes an analysis of socio-­
economic and socio-psychological factors for all terrorism offenders in the
Netherlands (N = 279). The study had two different control groups. One was a
sample of matched common criminals (N = 279), and the other a sample drawn
from the general population from official statistics (N = 6000). The study found that
compared to the general population, terrorism offenders were more likely to be first
or second generation immigrants, have criminal histories, and be unemployed or
recently experienced job loss.
Part II of this volume presents studies of recruitment to organized crime. As with
Part I on terrorism, Part II begins with a systematic review of the social, psychological, and economic factors relating to involvement and recruitment into organized
crime groups (OCGs), including mafias, drug trafficking organizations (DTOs),
gangs, and other criminal organizations. The review searched all possible relevant
studies indexed in selected databases and published in five languages (i.e. English,
Introduction
11
French, German, Italian, and Spanish), without limitations as to their year of publication or geographic origin. Starting from 48,731 potentially eligible records,
­integrated with experts’ suggestions, this systematic review includes 47 empirical
studies employing quantitative, qualitative, or mixed-methods approaches. The
findings show that social and economic factors are the most commonly reported
factors relating to recruitment into OCGs, while psychological factors are marginal.
Although factors are highly interrelated and shared across OCGs, their predominance varies across types of criminal organizations. For instance, individuals join
mafias and gangs attracted by strong group identity, whereas individuals enter DTOs
mainly for financial gain.
In the chapter “Socio-Economic Inequalities and Organized Crime: An
Empirical Analysis” Battisti and colleagues carry out a macro level study by
examining data spanning from 1985-2014. They investigate the relationship
between inequality and organized crime at the regional level, assessing the role of
social mobility in organized crime. The study found that higher levels of inequality are associated with greater organized crime development. The study also found
that lower socio-­economic mobility displays a robust association with organized
crime development.
Moving towards individual level factors, Savona and colleagues provide the first
analyses of the criminal careers of Italian Mafia members in the chapter “The
Criminal Careers of Italian Mafia Members”. The analysis is based on the unique
Proton Mafia Member dataset, provided by the Italian Ministry of Justice, with
information on all individuals who received a final conviction for mafia offences
since the 1980s. The PMM includes information on more than 11,000 individuals
and 182,000 offences. The study explores the criminal careers of mafia members
following a three-level approach, analyzing the macro, meso, and micro dimensions. At the macro level, Italian mafias’ members show different types of criminal
trajectories, with a significant portion of the sample exhibiting a late onset and late
persistence pattern. At the meso level, the four main types of mafias (the Sicilian
Cosa Nostra, the Neapolitan Camorra, the Calabrian ‘Ndrangheta, and the Apulian
mafias) report very similar traits although some distinctive patterns emerge. At the
micro level, there are differences in criminal careers between early- and lately
recruited members, with the former showing higher frequency of violent, volume
crime and the latter a more complex, white collar profile. A further exploration of
the PMM data shows an escalation in both the number and the seriousness of crimes
before joining the mafia, which subsequently stabilizes afterwards.
While previous studies have shown that criminal behaviour is associated with
antisocial, borderline, narcissistic, and other personality disorders (Butler et al.
2006; Fazel and Danesh 2002; Roberts and Coid 2010), these factors are rarely
assessed when it comes to organized crime offenders. Addressing this key gap in the
body of knowledge, Salvato and colleagues set out to investigate the psychological
profile of organized crime members in the chapter “Investigating the Psychological
Profile of Organized Crime Members”. They gave personality tests to 100 prisoners
(50 organized crime offenders and 50 common criminals) incarcerated in Italy. The
results showed that masochistic personality disorder was more likely to be found
among organized crime offenders than ordinary criminals.
12
D. Weisburd et al.
Moving to a different European context, Madarie and Kruisbergen explore organized crime in the Netherlands in the chapter “Traffickers in Transit: Analysing the
Logistics and Involvement Mechanisms of Organised Crime at Logistical Nodes in
the Netherlands”. The Netherlands functions as an important source and transit
country for international organised drug trafficking. This is in part due to its large
logistical nodes in the world economy, like the airport and the seaport. Although
knowledge on organised crime offenders engaging in transit crime has accumulated
recently, systematic analyses of offenders operating on logistical nodes are still rare.
Based on in-depth analyses of 16 cases of the Dutch Organised Crime Monitor, this
chapter explores how drug traffickers operate at logistical nodes, in particular airports. Furthermore, the occupational and social embeddedness of organised crime
activities and the personal motivations of drug traffickers to become and stay
involved are analysed. Although the main analyses focus on organised crime at airports (11 cases), organised crime at seaports (5 cases) is analysed as well in order to
compare the logistics and involvement mechanisms at the airport with those at the
seaport.
The results demonstrate that organised crime groups deploy mainly three types
of tactics to traffic drugs, namely defying, avoiding, and neutralising security
checks. Occupational embeddedness is manifested through several job-related factors. Autonomy, mobility, and the similarity between legitimate duties and criminal
activities facilitate discrete engagement in organised crime activities during work
time. Port employees are also attractive to organised crime groups because of their
job-related social capital and knowledge. Co-offenders are recruited through snowballing and are most often family members or former colleagues. Finally, social
bonds and money are both important in becoming and staying involved. Greed is
more often a reason to stay involved than a need for money. If members do want to
leave the group, threats of violence are not shunned.
In another study focussed on the Netherlands, Van der Geest and colleagues
explore the relationship between delinquent development, employment and income
in a large sample of Dutch organized crime offenders in the chapter “Delinquent
Development, Employment and Income in a Sample of Dutch Organized Crime
Offenders: Shape, Content, and Correlates of Delinquent Trajectories from Age 12
to 65”. The analyses provide an in depth exploration of offending and risk factors
over the life-course. The study found that on average, organized crime offenders’
peak offending occurred at ages slightly older than common criminals. The overwhelming majority of offenders had criminal histories preceding their first organized crime offence, and had multiple subsequent offences.
Overlap and Distinctive Patterns: What We Learned
As we noted at the outset PROTON not only sought to increase the knowledge base
about recruitment to terrorism and organized crime, it also sought to identify the
extent to which these areas of study would benefit from examination in the context
Introduction
13
of a single research program. Our chapter so far has documented the rich body of
new research our project has generated in both of these areas. We think that this is
the largest collection of essays focused on recruitment to organized crime and terrorism that has been published to date. And our studies provide important new
insights into the social, economic and psychological forces that lead to recruitment
to both terrorism of organized crime. But a key question we want to turn to before
concluding this introduction is whether this research encourages us to see recruitment as a general phenomenon for crime and terrorism research, or whether it suggests that little can be gained by their combination in a single area of study.
We want to emphasize again that the the studies in this volume reinforce what we
noted in our reviews of the literature to date. The study of risk factors for recruitment to terrorism and organized crime remains underdeveloped compared to similar
study pertaining to common criminality. Nonetheles, the studies included in this
volume provide important additions and contributions to the body of knowledge,
each exploring key topics through the use of a range of methods and unique samples.
A key conclusion of our project is that there is a substantial degree of overlap in
our findings across terrorism and organized crime. It is simply not right to say that
these areas of study are so distinct that each has little to inform in terms of the other.
In fact, the recruitment into both terrorist and organized crime groups is likely to
follow common pathways. Nor is it right to say that we can learn little from more
general studies of criminality and crime for understanding recruitment to organized
crime and terrorism. It is certainly the case that these crimes have important distinctions from each other and from common crimes. Nonetheless, the PROTON studies
show that factors relating to key criminological theoretical perspectives and empirical evidence pertaining to general criminality appear highly relevant for both organized crime and terrorism.
For example, socio-economic factors are key issues for understanding involvement in organized crime and in terrorism, as they are key factors in understanding
common crimes (Pratt and Cullen 2005). Nonetheless, it is important to point out
substantive differences that emerge from our studies. While socio-economic factors
are important for understanding recruitment to terrorism, objective economic deprivation appears to be less important for terrorism offenders than for common criminals. Rather, under-­employment and recent job-loss may be related to perceived
injustices, discrimination, and feelings of subjective deprivations and these feelings
are often key to understanding recruitment to terrorism. These differences highlight
the fact that while criminals and terrorists may arise from the same general pool of
the population, they differ in how economic factors influence their recruitment.
The studies also highlight the extent to which personality-trait related factors are
key for both organized crime recruitment and radicalization and recruitment to terrorism. Masochistic personality traits appear to be a risk factor for recruitment to
organized crime. Low self-control acts as an important risk factor for recruitment to
terrorism. These are interesting and important findings that suggest significant overlaps in recruitment. However, we want to emphasize again that the role of psychological factors remains under-researched, and this should be an area of significant
inquiry in future studies.
14
D. Weisburd et al.
Perhaps one of the most striking overlaps between terrorism and organized crime
recruitment is with respect to the role of criminal history for both organized crime
and terrorism offenders. The findings of the PROTON studies represent some of the
first explorations of this issue, and they suggest not only overlap between organized
crime and terrorism, but also that findings regarding common crime offenders are
relevant for understanding terrorism and organized crime.
The criminological literature has long found that prior criminality is one of the
strongest predictors of future offending, including repeat offending (recidivism),
across virtually every type of offence (e.g. Gendreau et al. 1996; Kyckelhahn and
Cooper 2017). Even in the case of white collar crime there is now strong evidence
of repeat offending (Weisburd et al. 2001; Benson and Simpson 2014). While the
radicalization literature has in recent years begun to focus on the role of prior criminality (e.g. Ljujic, van Prooijen and Weerman 2017; Rostami et al. 2018; Bakker
2006, 2011; Bakker and de Bont 2016; Weenink 2015; Basra and Neumann 2017;
van Leyenhorst and Andreas 2017; Boncio 2017; Heinke 2017; LaFree et al. 2018;
Smith 2018; Horgan et al. 2016), the evidence produced by the chapters in this volume shed new light on the extent of the issue. Prior criminality was found to be a
significant predictor for being a terrorism offender compared to the general population as well as non-violent extremists. It was also found to be a significant predictor
for recidivism to terrorism. Similarly, criminal expertise and criminal records were
found to be among the main factors for recruitment into organized crime.
Importantly, our studies also identified the importance of perceptions of legitimacy in preventing recruitment to terrorism. There is a growing body of literature in
criminology that focuses in on how police treat offenders and how that may have
strong influence on perceptions of police legitimacy (Mazerolle, Antrobus, Bennett
and Tyler 2013; Tyler 2017; for an opposing view see Nagin and Telep 2017).
Tankebe’s examination of perceived reactions to unjust police stops in the chapter
“Unintended Negative Outcomes of Counter-Terrorism Policing: Procedural (In)
Justice and Perceived Risk of Recruitment into Terrorism” shows that police behaviour can be a risk factor in recruitment to terrorism. This is an important finding, that
supports concerns about procedurally unjust police behaviour (Worden and
McLean 2017).
Our comments here suggest that the original motivating idea of PROTON was a
useful one for expanding knowledge of recruitment more generally. There are certainly differences that need to be recognized between organized crime and terrorism
recruitment. For example, the study of organized crime at air and sea ports described
in the chapter “Delinquent Development, Employment and Income in a Sample of
Dutch Organized Crime Offenders: Shape, Content, and Correlates of Delinquent
Trajectories from Age 12 to 65” emphasizes the difference between common crimes
and organized crime, and also points to distinctions with terrorism. But even here,
there are commonalities that are important. Recruitment for example, comes through
networks of friends or former workers. A failure to recognize such overlaps in our
understanding of these phenomena limits the development of serious academic
study. The examination of these problems in a single research program has allowed
Introduction
15
us to see the commonalities between these two types of criminal recruitment. We
think that this has enriched our research program, and its contributions to existing
knowledge.
Conclusions and Policy Recommendations
Our volume has added specific knowledge to the understanding of recruitment to
terrorism and organized crime. It also emphasizes as we noted in the previous section the extent to which core criminological ideas can be applied across these phenomena. Risk factors often overlap, despite some important distinctions, and the
mechanisms of recruitment are often similar. These studies suggest the importance
of key criminological perspectives for understanding recruitment to both organized
crime and terrorism, such as that of criminal careers (e.g. Kleemans and De Poot
2008; Horgan and Taylor 2011) and differential associations (e.g. Kleemans and van
de Bunt 1999; Reinares et al. 2017). Indeed, our volume highlights prior criminality,
and the role of social and work associations, as being important risk factors for
recruitment to both organized crime and terrorism.
These findings mean that in some sense traditional models of crime prevention
ought to be applicable and appropriate for dealing with both terrorism and organized crime. If we know that prior criminality is a strong predictor of involvement
in terrorism and recruitment it can point us to high risk populations and the importance of preventing radicalization in such populations. It can also point us to the
importance of prevention in prisons. At present, prevention programs generally
focus on reducing common crimes in the future. But the PROTON studies suggest
that prevention programs focusing on reducing radicalization or preventing involvement in organized crime should provide important foci for reducing recruitment.
Similarly, PROTON studies suggest the importance of socio-economic factors in
recruitment. Again, they identify specific populations where radicalization and
potential recruitment to organized crime are more common. But importantly they
also suggest how economic programs can be used to prevent recruitment. In the case
of terrorism, for example, it is clear that unemployment strongly increases risk of
recruitment. Accordingly, employment programs focused on individuals of high
risk for radicalization would seem to be an important potential prevention approach
for policy makers.
Reducing radicalization in turn may require better training of the police.
Reducing police activities that are perceived as unjust appears to be a salient
approach for reducing radicalization. In turn, interruption of social networks that
support organized crime recruitment would seem particularly important for reducing recruitment.
For the most part our studies have focused on the risk or protective factors that
increase or decrease recruitment, or the mechanisms behind recruitment. In this
sense our studies do not provide a direct set of policy proscriptions for reducing
16
D. Weisburd et al.
recruitment to organized crime or terrorism. At the same time, our volume provides
new and important evidence about the recruitment process and risk and protective
factors that encourage and discourage recruitment. This knowledge, we believe,
should inform practitioners and policy makers as they seek to develop evidence
based and effective programs to reduce recruitment to organized crime or terrorism.
References
Ahmed, S. (2016). Is history repeating itself: sentencing young American muslims in the war on
terror. Yale LJ, 126, 1520.
Albanese, J. S. (2014). Organized crime in our times. Routledge. New York
Allan, H. G., Glazzard, A., Jesperson, S. T., & Sneha Reddy Winterbotham, E. (2015). Drivers of
violent extremism: Hypotheses and literature review. London: Royal United Services Institute.
Armstrong, G., Derrick, D., Hienz, J., Ligon, G., & Southers, E. (2019). Characteristics of
Homegrown Violent Extremist Radicalization. Los Angeles: National Center for Risk and
Economic Analysis of Terrorism Events University of Southern California.
Aslam, F., Rafique, A., Salman, A., Kang, H. G., & Mohti, W. (2018). The impact of terrorism on
financial markets: Evidence from Asia. The Singapore Economic Review, 63(05), 1183–1204.
Basra, R., & Neumann, P. R. (2017). Crime as Jihad. CTC Sentinel, 10(9), 1–5.
Bassiouni, M. C. (1990). Effective national and international action against organized crime and
terrorist criminal activities. Emory International Law Review, 4, 9.
Benavides, A. D., Keyes, L. M., & Pulley, B. (2016). Understanding the recruitment methods and
socialization techniques of terror networks by comparing them to youth gangs: Similarities
and divergences. Countering Terrorist Recruitment in the Context of Armed Counter-Terrorism
Operations, 125, 40.
Benson, M. L., & Simpson, S. S. (2014). Understanding white-collar crime: An opportunity perspective. Routledge. New York.
Berrebi, C., & Klor, E. F. (2010). The impact of terrorism on the defence industry. Economica,
77(307), 518–543.
Bovenkerk, F., & Chakra, B. A. (2005). Terrorism and organized crime. Forum on Crime and
Society, 4, 3–16.
Blokland, A., Van Hout, L., van der Leest, W., & Soudijn, M. (2019). Not your average biker;
criminal careers of members of Dutch outlaw motorcycle gangs. Trends in Organized Crime,
22(1), 10–33.
Bakker, E. (2006). Jihadi terrorists in Europe: Their characteristics and the circumstances in which
they joined the jihad: An exploratory study. Netherlands Institute of International Relations
Clingendael.
Bakker, E. (2011). Characteristics of jihadi terrorists in Europe (2001–2009). In R. Coolsaet,
Jihadi Terrorism and the Radicalisation Challenge. European and American Experiences
(pp. 131–144). Aldershot: Ashgate.
Bakker, E., & de Bont, R. (2016). Belgian and Dutch jihadist foreign fighters (2012–2015):
Characteristics, motivations, and roles in the War in Syria and Iraq. Small Wars & Insurgencies,
27(5), 837–857.
Boncio, A. (2017). Italian Foreign terrorist fighters: a quantitative analysis of radicalization risk
factors. Count Terror Prev Radic Prot Cult Herit Role Human Factors Technol, 133, 40.
Bandiera, O. (2003). Land Reform, the Market for Protection, and the Origins of the Sicilian
Mafia: Theory and Evidence. Journal of Law, Economics, and Organization 19(1), 218–244.
Bondokji, N., Wilkinson, K., & Aghabi, L. (2017). Understanding Radicalisation: A Literature
Review of Models and Drivers. Ministry of Foreign Affairs of the Netherlands.
Introduction
17
Butler, T., Andrews, G., Allnutt, S., Sakashita, C., Smith, N. E., & Basson, J. (2006). Mental disorders in Australian prisoners: a comparison with a community sample. Australian & New
Zealand Journal of Psychiatry, 40(3), 272–276.
Calderoni, F. (2014). Mythical numbers and the proceeds of organised crime: Estimating mafia
proceeds in Italy. Global Crime, 15, 138–163.
Clarke, R. V. G., & Newman, G. R. (2006). Outsmarting the terrorists. Connecticut: Greenwood
Publishing Group.
Choudhury, T., & Fenwick, H. (2011). The impact of counter-terrorism measures on Muslim communities. International Review of Law, Computers & Technology, 25(3), 151–181.
Cornish, D. B., & Clarke, R. V. (2002). Analyzing organized crimes. Rational choice and criminal
behavior: Recent research and future challenges, 32, 41–63.
Decker, S. H., & Pyrooz, D. C. (2011). Gangs, terrorism, and radicalization. Journal of Strategic
Security, 4, 151–166.
Decker, S. H., & Pyrooz, D. C. (2014). Gangs: Another form of organized crime. In L. Paoli (Ed.),
The Oxford handbook of organized crime (pp. 270–287). Oxford: Oxford University Press.
Deflem, M. (2007). International police cooperation against terrorism: Interpol and Europol in
comparison. Understanding and Responding to Terrorism, 19, 17.
den Boer, M. (2015). Counter-Terrorism, Security and Intelligence in the EU: Governance
Challenges for Collection, Exchange and Analysis. Intelligence and National Security 30(2–3),
402–419.
Densley, J. A. (2012). Street gang recruitment: Signaling, screening, and selection. Social
Problems, 59(3), 301–321.
Desmond, E., & Hussain, N. (2017). Organized crime and terrorism. In J. D. Freilich & G. LaFree
(Eds.), The handbook of the criminology of terrorism (pp. 373–384). Hoboken: Wiley
Blackwell.
Dishman, C. (2001). Terrorism, crime, and transformation. Studies in Conflict & Terrorism, 24,
43–58.
Dishman, C. (2005). The leaderless Nexus: When crime and terror converge. Studies in Conflict &
Terrorism, 28, 237–252.
Dodd, V. (2019). Counter-terror chief says policing alone cannot beat extremism.
Retrieved from The Guardian: https://www.theguardian.com/uk-news/2019/aug/06/
counter-terrorism-chief-calls-for-greater-social-inclusion
Edens, J. F. (2006). Unresolved controversies concerning Psychopathy: Implications for clinical
and forensic decision making. Professional Psychology: Research and Practice 37(1), 59–65.
Fazel, S., & Danesh, J. (2002). Serious mental disorder in 23 000 prisoners: a systematic review of
62 surveys. The Lancet, 359(9306), 545–550.
Felbab-Brown, V. (2019). The crime-terror Nexus and its fallacies. In E. Chenoweth, R. English,
A. Gofas, & S. N. Kalyvas (Eds.), The Oxford handbook of terrorism (pp. 366–383). Oxford:
Oxford University Press.
Finelli, G. A. (2019). Slash, shoot, kill: Gang recruitment of children and the penalties gangs face.
Family Court Review, 57(2), 243–257.
Fleenor, P. (2003). Cigarette taxes, black markets, and crime. Policy Analysis, 468, 1–20.
Gendreau, P., Little, T., & Goggin, C. (1996). A meta-analysis of the predictors of adult offender
recidivism: What works! Criminology, 34(4), 575–608.
Gill, P. (2012). Terrorist violence and the contextual, facilitative and causal qualities of group-­
based behaviors. Aggression and Violent Behavior 17(6):565–574.
Gill, P., & Corner, E. (2017). There and back again: The study of mental disorder and terrorist
involvement. American Psychologist, 72(3):231.
Haddad, S. (2017). Accounting for Lebanese Muslims’ perspectives on the Islamic state (ISIS):
religious militancy, sectarianism and personal attributions. Defense & Security Analysis, 33(3),
242–262.
Hare, R. D. (1999). Psychopathy as a risk factor for violence. Psychiatric Quarterly, 70(3),
181–197.
18
D. Weisburd et al.
Hasisi, B., Perry, S., & Wolfowicz, M. (2019). Counter-terrorism effectiveness and human rights in
Israel. In International human rights and counter-terrorism (pp. 1–21).
Harcourt, B. E. (2006). Muslim Profiles Post-9/11: Is Racial Profiling an Effective Counterterrorist
Measure and Does it Violate the Right to Be Free from Discrimination?. U Chicago Law &
Economics, Olin Working Paper, (288).
Heinke, D. H. (2017). German foreign fighters in Syria and Iraq: The updated data and its implications. CTC Sentinel, 10(3), 17–22.
Horgan, J., Shortland, N., Abbasciano, S., & Walsh, S. (2016). Actions speak louder than words: A
behavioral analysis of 183 individuals convicted for terrorist offenses in the United States from
1995 to 2012. Journal of Forensic Sciences, 61(5), 1228–1237.
Horgan, J., & Taylor, M. (2011). Disengagement, de-radicalization, and the arc of terrorism:
Future directions for research. Jihadi terrorism and the radicalisation challenge: European and
American experiences, 173–186.
Horgan, J. (2005). The psychology of terrorism. London, England: Routledge.
Hutchinson, S., & O’malley, P. (2007). A crime–terror Nexus? Thinking on some of the links
between terrorism and criminality. Studies in Conflict & Terrorism, 30, 1095–1107.
Jones, O. D., Wagner, A. D., Faigman, D. L., & Raichle, M. E. (2013). Neuroscientists in court.
Nature Reviews Neuroscience, 14(10), 730–736.
Kleemans, E. R., Soudijn, M. R., & Weenink, A. W. (2013). Situational crime prevention and
cross-border crime. In Clarke, R. & Tilley, N. (Eds.) Situational prevention of organised crimes
(pp. 35–52). Willan: Portland.
Kleemans, E. R., & De Poot, C. J. (2008). Criminal careers in organized crime and social opportunity structure. European Journal of Criminology, 5(1), 69–98.
Kleemans, E. R., & Van de Bunt, H. G. (1999). The social embeddedness of organized crime.
Transnational Organized Crime, 5(1), 19–36.
Kleemans, E. R., & Van Koppen, M. V. (2014). Careers in organized crime. Encyclopedia of
Criminology and Criminal Justice 285–295.
Knight, S., Woodward, K., & Lancaster, G. L. (2017). Violent versus nonviolent actors: An empirical study of different types of extremism. Journal of Threat Assessment and Management,
4(4), 230.
Kruglanski, A., & Golec, A. (2004). Individual Motivations, The Group Process and Organizational
Strategies in Suicide Terrorism. In E. Milgrom, Suicide Missions and the Market for Martyrs, A
Multidisciplinary Approach. Princeton: Princeton University Press.
Kyckelhahn, T., & Cooper, T. (2017). The past predicts the future: Criminal history and recidivism
of federal offenders. Washington D.C.: United States Sentencing Commission.
LaFree, G., & Ackerman, G. (2009). The empirical study of terrorism: Social and legal research.
Annual Review of Law and Social Science, 5, 347–374.
LaFree, G., & Freilich, J. D. (Eds.). (2016). The handbook of the criminology of terrorism. John
Wiley & Sons.
LaFree, G., Jensen, M. A., James, P. A., & Safer‐Lichtenstein, A. (2018). Correlates of violent
political extremism in the United States. Criminology, 56(2), 233–268.
Lavezzi, A. M. (2008). Economic structure and vulnerability to organised crime: Evidence from
Sicily. Global Crime, 9, 198–220.
Lavezzi, A. M. (2014). Organised crime and the economy: A framework for policy prescriptions.
Global Crime, 15, 164–190.
Leith. (2002). Short-staffed Mafia makes a job offer you can’t refuse. Retrieved from The Daily
Telegraph: https://www.telegraph.co.uk/news/worldnews/northamerica/usa/1394814/Shortstaffed-Mafia-makes-a-job-offer-you-cant-refuse.html
Levi, M., & Maguire, M. (2004). Reducing and preventing organised crime: An evidence-based
critique. Crime, Law and Social Change, 41(5), 397–469.
Leukfeldt, E. R., & Kleemans, E. R. (2019). Cybercrime, money mules and situational crime
prevention: Recruitment, motives and involvement mechanisms. In Criminal networks and law
enforcement (pp. 75–89).
Introduction
19
Liem, M., van Buuren, J., de Roy van Zuijdewijn, J., Schönberger, H., & Bakker, E. (2018).
European lone actor terrorists versus “common” homicide offenders: An empirical analysis.
Homicide Studies, 22, 45–69.
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. SAGE publications, Inc.
Ljujic, V., van Prooijen, J. W., & Weerman, F. (2017). Beyond the crime-terror nexus: socio-­
economic status, violent crimes and terrorism. Journal of Criminological Research, Policy and
Practice, 3(3), 158–172.
Lum, C., Kennedy, L. W., & Sherley, A. (2006). Are counter-terrorism strategies effective? The
results of the Campbell systematic review on counter-terrorism evaluation research. Journal of
Experimental Criminology, 2(4), 489–516.
Lyman, M. D., & Potter, G. W. (2006). Organized crime (4th ed.). Upper Saddle River: Prentice
Hall.
Makarenko, T. (2004). The crime-terror continuum: tracing the interplay between transnational
organised crime and terrorism. Global Crime 6(1), 129–145.
Malik, O. F., Schat, A. C., Raziq, M. M., Shahzad, A., & Khan, M. (2018). Relationships between
perceived risk of terrorism, fear, and avoidance behaviors among Pakistani university students:
A multigroup study. Personality and Individual Differences, 124, 39–44.
Mazerolle, L., Antrobus, E., Bennett, S., & Tyler, T. R. (2013). Shaping citizen perceptions of
police legitimacy: A randomized field trial of procedural justice. Criminology, 51(1), 33–63.
McCauley, C., & Moskalenko, S. (2017). Understanding political radicalization: The two-­pyramids
model. American Psychologist, 72(3), 205.
Merari, A. (2004). Suicide terrorism. In R. I. Yufit, & D. Lester, Assessment, treatment, and prevention of suicidal behavior (pp. 431–454). Hoboken: John Wiley & Sons.
Mesch, G. S. (2000). Perceptions of risk, lifestyle activities, and fear of crime. Deviant Behavior,
21(1):47–62.
Mullins, S. (2009). Parallels between crime and terrorism: A social psychological perspective.
Studies in Conflict & Terrorism, 32(9), 811–830.
Nagin, D. S., & Telep, C. W. (2017). Response to “Procedural justice and policing: A rush to judgment?”. Annual review of Law and social Science, 13, 55–58.
Neumann, P. (2013). The trouble with radicalization. International Affairs 89(4), 873–893.
Neumann, P., & Kleinmann, S. (2013). How rigorous is radicalization research?. Democracy and
Security, 9(4), 360–382.
Ostrosky, F., Borja, K. C., Rebollar, C. R., & Galván, K. D. (2011). Neuropsychological profiles of members of organized crime and drug-traffic organizations, Research and Reports in
Forensic Medical Science, 2, 19–30.
Petrosino, A., Boruch, R. F., Farrington, D. P., Sherman, L. W., & Weisburd, D. (2003). Toward
evidence-­
based criminology and criminal justice: Systematic reviews, the Campbell
Collaboration, and the Crime and Justice Group. International Journal of Comparative
Criminology, 3(1), 42–61.
Pinotti, P. (2015). The economic costs of organised crime: Evidence from southern Italy. The
Economic Journal, 125, F203–F232.
Pluchinsky, D. A. (2008). Global jihadist recidivism: A red flag. Studies in Conflict & Terrorism,
31(3), 182–200.
Pratt, T. C., & Cullen, F. T. (2005). Assessing macro-level predictors and theories of crime: A
meta-analysis. Crime and Justice, 32, 373–450.
Pyrooz, D. C., LaFree, G., Decker, S. H., & James, P. A. (2018). Cut from the same cloth? A
comparative study of domestic extremists and gang members in the United States. Justice
Quarterly, 35, 1–32.
Rapoport, D. C. (1983). Fear and trembling: Terrorism in three religious traditions. American
Political Science Review, 78(3), 658–677.
Rapoport, D. C. (2004). The four waves of modern terrorism. In Attacking terrorism: Elements of
a grand strategy (pp. 46–73).
Reinares, F., Garcia-Calvo, C., & Vicente, A. (2017). Differential association explaining Jihadi
radicalization in Spain: A quantitative study. CTC Sentinel, 10(6), 29–34.
20
D. Weisburd et al.
Roberts, A. D., & Coid, J. W. (2010). Personality disorder and offending behaviour: Findings from
the national survey of male prisoners in England and Wales. The Journal of Forensic Psychiatry
& Psychology, 21(2), 221–237.
Rostami, A., Sturup, J., Mondani, H., Thevselius, P., Sarnecki, J., & Edling, C. (2018). The
Swedish Mujahideen: An exploratory study of 41 Swedish foreign fighters deceased in Iraq
and Syria. Studies in Conflict & Terrorism, 1–14.
Sageman, M. (2014). The Stagnation in Terrorism Research. Terrorism and Political Violence
26(4), 565–580.
Sarma, K. M. (2017). Risk assessment and the prevention of radicalization from nonviolence into
terrorism. American Psychologist 72(3), 278–288.
Savona, E. U., Riccardi, M., & Berlusconi, G. (Eds.). (2016). Organised crime in European businesses. London - New York: Routledge.
Savona, E. U., Kleiman, M. A. R., & Calderoni, F. (Eds.). (2017). Dual markets: Comparative
approaches to regulation. Springer.
Schelling, T. C. (1971). What is the business of organized crime?. The American Scholar, 643–652.
Schmid, A. P. (1996). The links between transnational organized crime and terrorist crimes.
Transnational Organized Crime, 2(4), 40–82.
Schmid, A. P. (2018). Revisiting the relationship between international terrorism and transnational organised crime 22 years later. Research Paper, The International Center for Counter-­
Terrorism, The Hague, 1–40.
Schuurman, B., & Bakker, E. (2016). Reintegrating jihadist extremists: evaluating a Dutch initiative, 2013–2014. Behavioral Sciences of Terrorism and Political Aggression, 8(1), 66–485.
Shaw, M., & Mahadevan, P. (2018). When terrorism and organized crime meet. Policy Perspectives,
6(7), 1–4.
Shelley, L. I. (1995). Transnational organized crime: An imminent threat to the nation-state?
Journal of International Affairs, 48, 463–489.
Smith, A. G. (2018). How radicalization to terrorism occurs in the United States: What research
sponsored by the National Institute of Justice tells us. US Department Of Justice, Office of
Justice Programs, National Institute of Justice.
Smith, R. G. (2014). Responding to organised crime through intervention in recruitment pathways.
Trends and Issues in Crime and Criminal Justice, 473, 1.
Stern, J. (2016). Radicalization to Extremism and Mobilization to Violence. The ANNALS of the
American Academy of Political and Social Science 668(1):102–117.
Taylor, M., & Horgan, J. (1999). Future developments of political terrorism in Europe. Terrorism
and Political Violence, 11(4), 83–93.
Thompson, N. (2019). ‘It’s a no-win scenario, either the police or the gang will get you’:
Young people and organised crime–vulnerable or criminal? Youth Justice. https://doi.
org/10.1177/1473225419843353.
Tyler, T. R. (2017). Can the police enhance their popular legitimacy through their conduct: Using
empirical research to inform law. University of Illinois Law Review, 2017, 1971.
Van Leyenhorst, M., & Andreas, A. (2017). Dutch Suspects of Terrorist Activity: A Study of Their
Biographical Backgrounds Based on Primary Sources. Journal for Deradicalization, (12),
309–344.
Veldhuis, T. M., & Kessels, E. J. (2013). Thinking before leaping: The need for more and structural
data analysis in detention and rehabilitation of extremist offenders. The Hague: International
Centre for Counter-Terrorism.
Victoroff, J. (2005). The mind of the terrorist: A review and critique of psychological approaches.
Journal of Conflict resolution, 49(1), 3–42.
Victoroff, J., Adelman, J. R., & Matthews, M. (2012). Psychological factors associated with support for suicide bombing in the Muslim diaspora. Political Psychology, 33(6), 791–809.
Vishnevetsky, M. (2009). Case study–youth gangs and terrorism in Chechnya: Recruitment, activities and networks. In The faces of terrorism: Multidisciplinary perspectives (pp. 151–168).
West Sussex: Wiley-Blackwell.
Introduction
21
Walsh, A., Beaver, K. M. (2009). Biosocial Criminology. In: Krohn M., Lizotte A., Hall G. (eds)
Handbook on Crime and Deviance. Handbooks of Sociology and Social Research. Springer,
New York, NY (pp. 79–101).
Weenink, A. W. (2015). Behavioral problems and disorders among radicals in police files.
Perspectives on Terrorism, 9(2), 17–33.
Weisburd, D., Farrington, D. P., & Gill, C. (2017). What works in crime prevention and rehabilitation: An assessment of systematic reviews. Criminology & Public Policy, 16(2), 415–449.
Weisburd, D., Waring, E., & Chayet, E. F. (2001). White-collar crime and criminal careers.
Cambridge: Cambridge University Press.
White, R. (1998). Curtailing youth: A critique of coercive crime prevention. Crime Prevention
Studies, 9, 117–137.
Williams, P. (1998). Terrorism and organized crime: Convergence, Nexus or transformation? FOA
Report on Terrorism, Stockholm, Defence Research Establishment, 69–92.
Wolfendale, J. (2007). Terrorism, security, and the threat of counterterrorism. Studies in Conflict
& Terrorism, 30(1), 75–92.
Worden, R. E., & McLean, S. J. (2017). Research on police legitimacy: The state of the art.
Policing: An International Journal, 40(3), 480–513.
Part I
Factors Affecting Radicalization and
Recruitment to Terrorism
What Is the State of the Quantitative
Literature on Risk Factors
for Radicalization and Recruitment
to Terrorism?
Michael Wolfowicz, Yael Litmanovitz, David Weisburd, and Badi Hasisi
Introduction
Terrorism research has become a field of increasing interest and importance over the
last few decades. Since the events of 9/11, the number of publications in the field
has increased exponentially. Despite the explosion in terrorism research, reviews of
the literature have consistently bemoaned the dearth of empirical evidence (Silke
2001, 2007, 2009; Sageman 2014). Among the most prominent explanations given
for the findings that as little as 3–5% of research is based on any form of empiricism, is the continued absence of definitional clarity and consensus. Scholars also
refer to the lack of original data, and a tendency for terrorism research to reuse the
same data sets many times over, as one of the primary limitations of the literature
(Schuurman and Eijkman 2013). Such criticisms were summarized in Sageman’s
(2014) infamous article on the ‘stagnation of terrorism research’.
However, others such as Schmid (2011) have had a more favorable and positive
view towards the state of the literature and have countered that the situation is one
that is much better than what others have painted. Indeed, it would seem that in
recent years the literature is heading in a more positive direction, with an increase
in the number of quantitative studies. According to Schuurman’s (2018) review of
3442 terrorism papers published between 2007–2016, 1.3% employed inferential
statistics and a further 5.8% used a combination of descriptive and inferential statistics. On top of this, 14.7% of studies employed descriptive statistics.
M. Wolfowicz · Y. Litmanovitz · B. Hasisi
Institute of Criminology, Faculty of Law, Hebrew University of Jerusalem, Jerusalem, Israel
e-mail: michael.wolfowicz@mail.huji.ac.il
D. Weisburd (*)
Institute of Criminology, Faculty of Law, Hebrew University of Jerusalem, Jerusalem, Israel
Department of Criminology, Law and Society, George Mason University,
Fairfax, Virginia, USA
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_2
25
26
M. Wolfowicz et al.
However, these reviews do little to inform us about the state of the literature on
the different sub-fields that make up the terrorism literature (Reid and Chen 2007).1
Radicalization, one such sub-field that has dominated much of the terrorism discourse in recent years, remains a relatively young topic and perspective (Neumann
2013; Neumann and Kleinmann 2013; Schmid 2013). While much has been said
about the broader terrorism literature, little is known about the state of the sub-­
literature on radicalization. Neumann and Kleinmann (2013) believe that the definitional issue has also inhibited the development of radicalization research. In their
review of the rigour of radicalization research, Neumann and Kleinmann (2013)
find that 74%of the studies used qualitative methodologies, mostly case studies and
narrative techniques, while 20% used quantitative methods. While only 26% of the
qualitative studies demonstrated a high level of methodological rigor, the authors
assessed that a whole 94% of the quantitative studies were of high methodological
rigor. As such, the state of the literature on radicalization appears to be somewhat
more positive relative to Schuurman’s (2018) findings on the broader terrorism literature. Moreover, it would seem that quantitative research on radicalization
accounts for a significant proportion of the terrorism literature’s quantitative
publications.
While a number of other systematic reviews have sought to explore radicalization as a specific subfield of terrorism research, these reviews have been overly
broad in their scope. Some reviews have even readily admitted that this approach
has impeded their ability to achieving their own stated objectives (Christmann
2012). While these review provide important contributions, they are unable to
answer specific questions, such as; What does the quantitative evidence say? What
are the risk factors for radicalization? These are core topics of importance to radicalization research, especially given the findings of a recent systematic review on
risk assessment tools which found them to be of generally poor quality and lacking
an evidentiary basis (Scarcella et al. 2016). Indeed, systematic reviews are meant to
address focussed and narrow research questions, unlike traditional reviews which
explore topic areas (Gough et al. 2012; Khan et al. 2003).
This review seeks to address some of the issues that have hindered previous ones
by in asking the specific and focussed research question of what are the risk factors
of different radicalization outcomes in democratic countries as derived from quantitative studies? In asking this question we also seek to detail the qualities and characteristics of the quantitative literature in order to draw comparisons with the
broader literature. In order to address this question and contribute to important gaps
in the body of knowledge, this chapter is split into three sections. First, we conduct
a review of prior reviews on radicalization and in doing so identify the primary limitations and methodological issues that characterize these studies. On this basis, the
second section outlines the theoretical basis for our systematic review’s methodolOther subfields include but are not limited to: counter-terrorism, counter/de-radicalization and
disengagement (Dalgaard-Nielsen 2018; Windisch et al. 2017), lone-wolf terrorism (Spaaij and
Hamm 2015), geography of terrorism (Farrell et al. 2019), cyber-terrorism (Ducol et al. 2016), and
foreign fighters (Hegghammer 2010).
1
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
27
ogy, specifically with respect to the selection of inclusion criteria. In the third section, we present the results from our systematic review of the quantitative literature
on radicalization. We discuss the characteristics of the identified literature, ­including
its quality and rigour, and identify the most commonly examined risk factors examined in this body of evidence.
A Review of Prior Reviews
While terrorism research has been conducted by academics since the 1960’s, it
never really developed into an independent field of science. Rather, scholars and
researchers from a wide array of fields have at times contributed to the study of terrorism. Over the years, terrorism research has gone through waves of popularity
across different disciplines, including but not limited to, political science, sociology, psychology, and history (Reid 1997). In more recent years, scholars from fields
such as psychology and criminology have increasingly devoted their attention to
terrorism research (Neumann and Kleinmann 2013; LaFree and Freilich 2016;
Kruglanski et al. 2009). It may even be the case that today, scholars from these disciplines dominate the field of terrorism research. Despite the multi-disciplinary
attention that terrorism has received as a topic of research, and the large volume of
publications produced in the wake of 9/11, reviews of the literature have consistently painted a bleak portrait of literature’s state.
The first review that sought to provide an overview of the state of the terrorism
literature was conducted by Schmid and Jongman (1983) who identified that the
terrorism literature was quite weak. A few years later they conducted an updated
review in which they stated that there were probably few other fields where so much
has been written on the basis of so little research (Schmid and Jongman 1988).
Almost a decade later, Reid (1997) reviewed over 1100 items published between
1960 and 1990 and found that the research overwhelmingly tended to examine
small insurgent groups. She also found that many studies simply re-categorized or
re-analyzed the same data or sets of terrorism incidents that prior studies had used.
Starting in 2001, Andrew Silke began to publish the first of several reviews of the
terrorism literature. Following Silke’s findings from 2001 he found that there was
little notable change in the quality and trajectory of the literature since Schmid and
Jongman. In 2004, Silke found that only 3% of terrorism related studies used inferential statistics and only 15% used some form of descriptive statistics. In 2008 he
found more positive developments, with an increased use of descriptive and inferential statistics in publications which had grown exponentially since 2001. However,
he makes a caveat that compared to other fields the findings are still quite poor. and
only in 2001 was Andrew Silke’s review published. Silke has since published a
number of reviews, every few years. In his 2007 review examining whether the post
9/11 period had impacted the quality of the research.
While these reviews provided important overviews of the literature, they still
were unable to deal with synthesizing the evidence. As Ranstorp (2006) noted, “the
28
M. Wolfowicz et al.
relative absence of any reflective state-of-the-art reviews of what the field has
achieved” (p.3), represents a gap in the current knowledge. However, starting with
Lum et al. (2006) systematic review of the effectiveness of counter-terrorism
strategies, a few systematic reviews have been published over the years.
­
Unfortunately, with the exception of this review, which also found a very small
proportion of the literature to be empirically based, let alone quantitative, most studies can only be characterized as being narrative reviews. While this could simply a
reflection of the nature of the literatures being examined, it may also be related to
the methodologies used by the reviewers, especially related to the development of
research questions.
Taking a broad approach to the development of systematic review research questions has been a characteristic of reviews related to radicalization. For example,
Christmann (2012) examined the topic of Preventing religious radicalization and
violent extremism. In the section on individual risk factors (pp. 31–34), the author is
only able to provide a narrative review of the evidence contained in the two dozen
or so studies included in this part of the review. The author notes the inconsistencies
and great heterogeneity identified in the studies. The author does stress two important points here however. First, that no single factor can or will explain radicalization, rather, it is the combined and cumulative effect of risk factors that increases the
likelihood of radicalization. Secondly, risk factors are most relevant to creating the
initial openings to radicalization which increase the likelihood than an individual
will come into contact with other radicalizing agents. These radicalizing agents he
believes, become more important in the subsequent stages of radicalization leading
to violent extremism.
This view is shared by the authors of another review on the same topic which was
carried out on behalf of the International Centre for the Prevention of Crime (ICPC)
by Madriaza and Ponsot (2015). Entitled “Preventing radicalization: A systematic
review”, the authors explain that the main idea behind the prevention of radicalization is that: “the social prevention of radicalization aims to limit the development of
risk factors and strengthen factors that protect against such a process.” (p. 8). The
review includes 483 items but little information is given as to how they were
selected. The authors note that their inclusion criteria limited itself to items published between 1 January 2005 and 1 June 2015 in either English or French.
However, certain items which the authors considered to be of great importance and
which fell outside of these criteria (e.g. Sageman 2004) were included. The authors
state that they used the following working definition of radicalization to lead their
inclusion criteria: “radicalization leading to violence”: [translation] “the process
whereby an individual or group adopts a violent form of action, directly related to a
politically, socially or religiously motivated extremist ideology that challenges the
established political, social or cultural order” (Khosrokhavar 2014, pp. 8–9).
However, it is not clear if this means that only studies whose dependent variables
met this definition were included or if it was used more broadly as criteria for inclusion by topic.
In section “A Review of Prior Reviews” of this review (pp. 27–44), the authors
delve into the factors of radicalization identified. While Internet use, propaganda,
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
29
and prison radicalization are discussed in isolation, the remaining 32 factors are
split up into vote-counting matrices where they are divided between Islamic and
right-wing radicalization and coded as either having consensus support or being
characterized by either insufficient or contradictory evidence. The first matrix lists
14 individual-level factors: Socioeconomic level, Employment, Education, Criminal
behaviour, Personal crisis, Age (being young), Gender (men), Negative experiences,
Military training, National identity (exclusion and xenophobia), Nationals/immigrants, Search for identity, Religion, and Mental health. Another 7 factors are presented in a separate matrix on relational/group based factors: Internal group
cohesion, Family, Charismatic leaders, Relationship with other adults, Personal networks, Resources, and Intergenerational transmission of radical ideas. Another 3
factors are examined as community related factors: Isolated community, Cultural
conflicts, and Radicalization subculture. Lastly, 8 factors are listed under the heading of ‘macro and exo-systemic factors‘: Traditional gender identity, Weak or unstable nations, Fast-paced modernization, Other macro-political factors, Cultural
heterogeneity, Social integration, Conflicts in Muslim countries, Authoritarian societies. For the Islamic group, half of the factors were found to be characterized by
insufficient or contradictory evidence, and 65% for right-wing radicalization.
Another review by McGilloway et al. (2015) set out to identify and synthesize
empirical studies only. This review included 17 papers, the majority of which were
qualitative (N = 13). The review identified a limited number of factors: Identity,
deprived neighborhood, discriminatory counter-radicalization polices (integration
and socio-economic approaches), socio-economic status, gender, age, citizenship,
convert to Islam, mental health, stressful life events, social contacts, racial discrimination, grievances over foreign policy, ideology, in-group superiority/ criticism.
Among these 15 factors, the evidence was presented as being quite mixed. The
authors summarize that the only consistent finding was that young, relatively well
integrated males are at the greatest risk for radicalization. Overall, the authors
remind us that: “Many theories are available, yet empirical research is lacking” (p.49).
Another review, published under the auspices of the UK’s Home Office, and carried out by Munton et al. (2011), was a scoping review on Understanding vulnerability and resilience in individuals to the influence of Al Qaida violent extremism.
One of the advantages of this review’s approach is that it’s title indicates that it is
examining a very specific research question, which is what systematic reviews
ought to do. Unfortunately, the review digresses somewhat from its focus, examining as a secondary research question, “What factors enable vulnerable individuals to
resist the influence of AQ-influenced violent extremism?”. While this is certainly
related to the primary research question, it can be seen as a different topic. Following
this, the review includes secondary findings from another review on “other types of
violent activity“. While the focus on “other (non-AQ-influenced) types of terror
activity” is still closely related to the topic and outcome of interest, it also includes
more distantly related phenomena such as; animal rights activism, cults, gangs, and
youth crime. That the review finds a lack of empirical evidence, and acknowledges
the difficulties in synthesizing the findings, may have been impacted by other meth-
30
M. Wolfowicz et al.
odological decisions, such as the review’s inclusion of disparate contexts, from
Western countries to Chechneya and Palestinian terrorists. The primary review consisted of 39 included studies using a variety of methodologies: controlled trials
(N = 2), interviews (N = 15), survey interview (N = 5), observational study (N = 1),
and secondary analysis of autobiographies or biographies, extremist pamphlets and
other communications (N = 16). The authors categorized the factors identified in
these studies into 13 categories, some of which covered multiple factors (e.g. age,
sex and marital status were grouped together), while others examined individual
factors (e.g. religious beliefs). The limited number of factors identified in this review
provided little innovation or new knowledge.
Focusing on the “subfield of research” of the ‘root causes’ of non-suicide terrorism, Campana and Lapointe’s (2012) review included 48 studies published between
2000 and 2009. While this review also appeared to have a more focussed scope and
research question, the researchers identified that the included studies could be
divided into four categories based on their dependent variables: (1) the number of
terrorist attacks in a country (or countries), (2) the number of foreign/transnational
terrorist attacks, (3) the number of terrorist groups operating in a country (or countries), and (4) structural factors explaining individual participation in terrorism
groups. Due to the difficulties in classifying and categorizing the independent variables, the researchers used a vote counting method to arrange the factors across
these four outcomes. With respect to the fourth dependent variable, and relevant to
the study of risk factors of radicalization, the researchers found evidence for: political freedom, civil liberties, state repression, macro socio-economic conditions,
macro-economic inequalities, individual poverty, individual-level education.
However, as reflected by the first three constructs of these dependent variables, little
can be learned about radicalization from them. Also, as the researchers bemoan, the
disparate nature of these dependent variables leads to an inability to synthesize the
evidence in a more coherent or useful way. Even this study, which presented a much
more focussed research question than the others mentioned above, suffered from the
lack of well defined inclusion criteria regarding which outcomes of interest it would
be limited to.
With regards to radicalization, while there are certainly many different conceptions of what it constitutes (Neumann and Kleinmann 2013), comparing ‘apples and
oranges’ is problematic in systematic review research. While this term generally
refers more to the quality of studies included in a meta-analysis, it is also important
that any systematic review limits itself to examining the same outcome or dependent
variable. In all of the above noted examples, the lack of a well-defined and narrow
inclusion criteria regarding the outcome of interest led to the comparing of studies
whose dependent variables were starkly different. For example, the review included
and compared studies which assessed radicalization based on ‘support for
terrorism‘(Tausch et al. 2009), ‘opposition to the West’ (McCauley 2012), and
Muslims perceptions of the ‘war on terrorism’ (McCauley and Scheckter 2008).
While acknowledging the pitfalls of a broad research question and inclusion criteria, Christmann (2012) takes such an approach anyhow. In taking such an approach,
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
31
these studies fall short of systematic review standards, which include the need of
asking focused, specific questions, and developing narrow and limiting inclusion
criteria. Sometimes taking a broader approach and trying to include everything
leads to us knowing less. As Schmid and Jongman (1988) noted three decades ago,
terrorism research exceeds every other field in having so much written about the
topic on the basis of such little empiricism.
imiting Inclusion Criteria in Radicalization Research
L
Synthesis
While achieving definitional consensus ought to be an objective of the field of radicalization research, it is not necessary to reach such a consensus in order to move
the field forward (Neumann and Kleinmann 2013). Individual studies each choose
their own operational definitions and the results of these studies either support or
detract from the validity of the conceptualization and operationalization. While
there is wide ranging support for the notion that ‘support for terrorism’ is an outcome which represents radicalization well, other outcomes, even if framed as proxies for radicalization, seem more distantly related, even if they satisfy some
definitions of radicalization, radicalism, and extremism.
For example, Sedgwick's (2010) conception of radicalization includes the opposition to western, democratic values, which follows early conceptions of political
radicalism as developed in the field of political science and the study of social
movements. As such, studies who frame their analyses of radicalization by exploring opposition to the West (McCauley 2012) may satisfy this definition of radicalization. However, other proxies for this approach to radicalization, such as voting
for populist parties, and authoritarian personality are arguably more problematic
and more difficult to justify. As Mudde (2004) has previously explained, the suggestion that populist parties or their voters oppose western or democratic values is
highly subjective. In fact, these parties and their supporters often run on platforms
in which they claim to be the vanguards of such values. With respect to the use of
right-wing authoritarian personality (RWA) or fundamentalism as proxies for radicalism, Koopmans (2015) believes that these make for poor proxies since they lack
reference to the actual use of violence. Moreover, these personality traits are not
“negative” in the same way that the type of radicalization we are usually referring
to is. While these personality traits may be risk factors for radicalization, they are
not good proxies for it (Stankov et al. 2018; McCleary et al. 2011; Koopmans 2015).
While examining conceptions of radicalization such as opposition to democracy are certainly an important part of the conversation on radicalization, at its
core, radicalization is the process of the development of the ideas, beliefs, or
opinions which could lead to acts of radical violence, including terrorism
(Neumann and Kleinmann 2013; EU 2014). But what types of beliefs should we
be focusing on remain ill-­defined. As Christmann (2012) notes in his systematic
32
M. Wolfowicz et al.
review, “Where the research question is broad-ranging or where the key terms are
vague or ill-defined, the ability of a systematic review to produce a convincing
answer is reduced”. Since “the ‘process of radicalisation’ remains ill-defined”, it
is difficult to collate, summarize, and synthesize the evidence according to established standards of systematic reviews.
While radicalization may be ill-defined, social-psychological theories that examine the attitudinal or cognitive antecedents of behaviors tell us that the level of
specificity is what determines which attitudes should be used as proxies for the
outcome of interest. On this basis, Moskalenko and McCauley (2009) cite the
Theory of Planned Behavior model in their work that developed into the Two-­
Pyramid of Radicalization’s cognitive radicalization outcomes (McCauley and
Moskalenko 2017). The Theory of Planned Behavior model is based on (1) attitudes, (2) intentions, and (3) behaviors. According to the model, attitudes lead to
intentions which in turn lead to behaviors. However, in order for an attitude to have
satisfactory level of predictive quality, it must meet a high level of specificity with
reference to the behavioral outcome of interest. Indeed, the Two-Pyramid Model’s
cognitive radicalization outcomes are: Support for terrorism, justification of terrorism, and a belief in a personal moral obligation to participate in terrorism. The first
two outcomes are attitudinal antecedents closely related to the behavioral outcome
of interest, whereas the third is the intentional dimension.
As has been pointed out in the literature, radical attitudes rarely lead to radical behaviors. A very small minority of individuals who hold radical attitudes
will ever turn to violence. In addition, there are cases of terrorists who were not
overly radicalized. Failing to be able to explain these two observations has been
a primary criticism of popular radicalization models and theories. However,
when extending the Theory of Planned Behavior to criminal behaviors, Beck
and Ajzen (1991) explain that the cumulative and interactive weight of risk factors are what determine when attitudes will translate into behaviors. The absence
of risk factors, or the presence of protective factors, explains why many with
favorable attitudes, or even expressed intentions, will never engage in the undesirable behavior. Criminologists have already found good evidence to support
this theory in analyzing how and when criminal attitudes predict criminal
behaviors (Folk et al. 2018).
As such, this framework fits exceptionally well into the study of risk factors
for cognitive and behavioral outcomes of radicalization. For the purposes of this
review, we limit our focus to studies that achieve a satisfactory level of specificity and examine cognitive radicalization by assessing support for or justification
of violent extremism (attitudes), intentions to engage in violent extremism
(intentions), or actual involvement in violent extremism (behaviors). By limiting our inclusion criteria, we are not trying to create a situation in which we
impose our chosen definitions on others, or by which we seek to ignore or
underplay the existence of additional, alternative, or event competing frameworks. Rather, we seek to achieve the highest level of specificity and thereby
enable the collation of studies that are measuring the same thing, which enables
a synthesis of the same.
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
33
The Current Study
As noted above, previous studies’ decisions to include all research, or to include
even all types of empirical research, represent a methodological limitation that has
hindered their ability to identify and synthesize the evidence, as reflected in authors’
own statements and their results. Moreover, whereas previous reviews have generally been interested in radicalization in western countries, they include studies from
non-western countries. While studies from non-western countries can certainly provide important insights, the mixing of studies from such varied contexts is invariably a source of the heterogeneity that these prior reviews have identified. In this
study we limit our scope to the identification and synthesis of quantitative studies
that examine ‘risk factors’ (or predictors) of a well defined and highly specific conception of radicalization. We further limit our inclusion to only those studies whose
populations and contexts are in democratic, high-income countries since low and
high income countries are known to differ significantly and combining them introduces an additional level of heterogeneity. There is therefore good methodological
reason for systematic reviews to examine either high or low income countries separately (Shenderovich et al. 2016; Higginson et al. 2014, 2018; Murray et al. 2018).
This review therefore set out to assess the quality of this literature and identify
which factors have been examined using rigorous quantitative methods. As described
in the above literature review, while qualitative research is certainly important, it has
widely been surveyed and reviewed. While some reviews have focussed exclusively
on empirical studies, no study has focussed exclusively on the quantitative literature. This review sought to include studies which employed appropriate statistical
methods for isolating the effects of individual predictors that are associated with an
increased likelihood or risk that someone will either adhere to radical attitudes (cognitive radicalization), and/or engage in radical behaviors (behavioral radicalization).2
The inclusion and exclusion criteria were pre-determined in a review protocol that
was registered with the Campbell Collaboration and which included parameters
pertaining to each component of a study: population/ measurement of outcome
(dependent variable), outcomes, context of the study, and research design (sample
size, analytic strategy, level of analysis).
The first set of criteria applied to assessing whether studies met the inclusion
criteria related to the population of interest; individuals with radical attitudes, radical intentions, or radical behaviors. In order to identify risk factors for radical
behaviors, a sample of non-violent radicals (LaFree and Freilich 2016), or the non-­
violent general population is needed to serve as a suitable comparison group
(Berrebi 2007; Krueger 2008). Accordingly, for radical attitudes, a comparison
group made up of those without radical attitudes was needed (McCauley and
Moskalenko 2017; Moskalenko and McCauley 2009) in order to provide the type of
variation on the dependent variable which is needed in observational studies
(Higginson et al. 2014).
2
A description of the methods used in the included studies is provided below.
34
M. Wolfowicz et al.
The second set of criteria pertained primarily to ensuring that the dependent variable met the inclusion criteria and assessed on one of the three dependent variables
of interest: (1) Justification/support of radical violence, (2) Personal moral obligation/willingness/intentions towards radical violence, (3) Involvement in radical violence (including terrorism). As such, studies falling outside of this scope, such as
comparative studies of terrorist offenders with ordinary criminal offenders (e.g.
Lyons and Harbinson 1986; Liem et al. 2018), or different types of terrorists, such
as religious and right-wing (e.g. Smith and Morgan 1994) were excluded.
Additionally, studies examining macro-level predictors of terrorism events, number
of terrorism groups, or recruitment flows were excluded on this basis as well.
ontext: Separating Between Democratic and Non-democratic
C
Countries
Systematic reviews often separate between limit their focus to specific types of
countries. Differences in political, social, economic and cultural contexts, which are
known to effect factors of radicalization (Brockhoff et al. 2015; Zhirkov et al. 2014;
Enders and Hoover 2012), mean that there is good methodological justification for
focusing on particular types of countries (Shenderovich et al. 2016; Higginson et al.
2014; Murray et al. 2018). Since there is no established preference as to whether
reviews should separate by income or regime type, this review bridged both
approaches by limiting inclusion to democratic countries who are also OECD countries. As such, the Democracy Index was cross matched with the OECD countries,
resulting in the inclusion of all of the 36 OECD countries with the only exclusion
being Turkey.
Research Designs
Systematic reviews usually aspire to include mostly experimental or quasi-­
experimental studies, in order to synthesis findings from the top of the hierarchy of
evidence. Given that the literature is known to be bereft of randomized control trials
(Laycock 2012; Victoroff 2005), the review sought to include only high-quality
cross-sectional studies, and the rare longitudinal study or quasi-experiment. When
dealing with literatures of primarily observational studies, inclusion criteria need to
be adjusted in order to avoid to ensure the rigor of included studies and avoid combining high and low quality studies, thereby biasing results (Shenderovich et al.
2016; Murray et al. 2018). As such, inclusion was limited to cross-sectional studies
which included: (1) samples >N = 50, and (2) employed appropriate statistical analysis and controls; such as calculating either bivariate or multivariate correlations
(e.g. Pearson’s r or regression coefficients). For longitudinal and experimental or
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
35
quasi-experimental studies, smaller samples would be considered for where they
met the other criteria.
Outcomes
While this review focusses on studies examining individual-level risk factors, in
order to avoid bias, we excluded references to putative factors from the searches
(e.g. education, employment, collective relative deprivation etc.). This decision was
made to ensure that no pre-determinations would be made as to what the risk factors
of radicalization are, particularly those which are known to be popular in the literature, thereby allowing us to answer our first primary research question without bias
(Hammerstrøm et al. 2009; Shenderovich et al. 2016).
Search Strategy
After conducting multiple pilot searches and consulting with librarians and academics with expertise in systematic searches in social and behavioral sciences, the final
search terms, search strings, and databases were selected. Searches were performed
on: The Campbell Collaboration library, ISI Science, PsychINFO, PubMed, SSRN,
SCO, Sociological Abstracts, Bibsys, START and journals such as Journal of de-­
radicalization, and Perspectives on Terrorism.3 Additionally, we contacted a number of experts from the US, EU and Israel to seek out unpublished studies and limit
the effects of publication bias (Higgins and Green 2011).
Results
Our searches resulted in the retrieval of 10,449 items. After removal of duplicates
we were left with 7892 original items which were screened based on their titles and
abstracts. A total of 7487 items were deemed to not meet the inclusion criteria, primarily based on their topic evidently being outside the scope of the review. The
remaining 405 items went through a secondary screening which included a closer
reading of the paper’s methodology sections. This screening led to the removal of
253 items, again primarily based on the topic being outside the scope of the review.
The remaining 152 items were read in their entirety by the senior researchers who
identified 50 studies that met all of the pre-determined inclusion criteria. In a­ ddition,
While most non-English studies are indexed in English, supplementary searches were also conducted in German and Dutch. General searches were also conducted on the Google scholar and
search engine to try and identify grey literature.
3
36
M. Wolfowicz et al.
Fig. 1 Systematic search results
our project partners and other experts who we contacted provided additional studies, 3 of which met the inclusion criteria. As such the final inclusion consisted of 53
studies.
Publications by Year
The results of the systematic searches identified a relatively large number of studies
that met the inclusion criteria. Overall, the studies were published between 2007
and 2018, with the largest number of studies (N = 10) being published in 2016. Only
one year—2008—had a single publication. The trend in publications shows a clear
uptick in recent years. While 2017 had fewer studies (N = 7) than 2016, and only
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
37
12
10
8
6
4
2
0
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Fig. 2 Publications by year
N = 6 studies were identified for 2018, it is possible that late 2018 publications,
some of which may not appear until 2019, will lead to an increased number (Fig. 2).
Some of the studies (N = 6) contain more than one individual study and as such
the total number of studies included from the 53 publications is 59.
The overall majority of the studies were found to have been published in English
although 4 German, 4 Dutch and 1 Spanish language study were identified as meeting the inclusion criteria. Of the German studies, two of them were written by the
same primary author (Dirk Baier). Among the Dutch studies, one study (Van den
Bos et al. 2009) used the same dataset that the same group of scholars would use in
a number of later English Publications (Doosje et al. 2012, 2013; Macdougall
et al. 2018).
Context
The majority of the studies focused on a European context. Of these studies, 33
focused on a specific country, whereas 4 focused on more than one country. The EU
country with the largest number of studies was the Netherlands which had 9 studies.
The second largest number of studies by country was for the UK which had 8 studies. Another 6 studies focused on Belgium, however these studies were all based on
the same original dataset. Another 4 studies focused on Germany, each employing
unique datasets. Another 2 studies focused on Poland, while Denmark, Spain,
Norway, and Switzerland were the focus of single studies each. Of the studies
­focusing on multiple countries, one study included all countries in the European
Values Survey (Fig. 3).
The next largest group of studies was from the US, with 9 studies. Another 2
studies focused on Israel, while 1 study was focused on Australia and another on
Canada. Two additional studies included samples from more than one region. One
of these included a sample of terrorists from multiple democratic countries, primarily from the EU and the US.
38
M. Wolfowicz et al.
Fig. 3 Number of studies
by region/country
1
1
2
2
8
39
Australia
Canada
Israel
Fig. 4 Number of studies
by ideology examined
Mixed
US
EU
6
17
30
Islam
Mixed
RWE
Populations
The largest proportion of the studies examined radical Islamic ideology as the focus
of the study (N = 30). However, a large number of studies (N = 17) examined mixed
samples (e.g. Muslim and non-Muslim, or RWE and Islamic, or mixed with no distinction). Another N = 7 studies examine RWE explicitly. One publication included
separate studies in which 1 examined single-issue separatist radicalization and this
is included among the 17 mixed studies (Fig. 4).
Authorship
The overwhelming majority of the studies identified had more than one author, with
only 9 studies having been published by a lone author. Among the studies published
by single authors, one was a thesis paper and another was a research paper. Among
the studies published by multiple authors, 31 displayed inter-institution cooperation, whereas 12 studies were published by authors from the same institution.
In terms of the fields from which authorship was derived, based on the primary
authorship the most frequent field was psychology, which had 14 studies. Among
these studies, 3 of them came from departments which were identified as being
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
39
Authorship by field
Terrorism/social sciences
Sociology
Social sciences
Public health
Psychology
Political science
Education/social sciences
Economics
Comunications
0
2
4
6
8
10
12
Fig. 5 Authorship by field
political psychology departments specifically. There were an additional 4 studies
from psychiatry. Among the remaining studies, 10 were from the field of criminology, and there were an additional two studies from departments listed as s­ ocial-­work/
criminology departments. A further 6 studies came from political science, 3 from
economics. There were a total of 5 studies from sociology, one of which was from
sociology/economics. Another two studies were from authors from the field of education, one of which being from a social sciences and education department, and a
further two studies from social-sciences. There were also two studies from the field
of public health. Lastly, there was a single study from a social-sciences/terrorism
department, a single study from law, and another single study from communications
(Fig. 5).
Dependent Variable
Overall, the majority of the studies examined cognitive radicalization in the form of
attitudes. Only 2 studies examined radical intentions as their sole dependent, while
another 10 studies examined radical intentions as a co-dependent variable with radical attitudes. Whilst some studies used more established instruments, such as the
Activism-Radicalism-Intentions-Scales (ARIS), or SyFor, the majority of studies
used dichotomous variables assessing support or justification for terrorism, specific
types of terrorism (e.g. suicide bombings), or radical violence4 (See Fig. 6).
Among the 12 studies that examined radical behaviors, 5 were based on self-­
reports of radical behaviors which can be characterized only as sub-terroristic radi-
Radical violence differed from justification of terrorism in that most studies referred to the use of
illegal violence to cause political or social change, or the use of illegal violence and actions to
protest government policies.
4
40
M. Wolfowicz et al.
16
14
12
10
8
6
4
2
0
Fig. 6 Distribution of dependent variables
cal violence, whereas the remainder examined terroristic radical behaviors. Of these
studies, 3 compared violent terrorists with non-violent terrorists, whereas 4 compared terrorists with a sample of the normative, general population.
Data Sources
A total of 74% of the included studies used original data, primarily derived from
surveys. Secondary data was used by 3 of these studies only for the purposes of
providing a comparison groups. Of the 12 studies that employed exclusively secondary data, 41% of them (N = 5) included data from different versions of the PEW
reports. Other secondary data sources included the European Values Surveys, of
which different sets were used in 2 studies, and the PIRUS database of US based
extremist offenders which was used in 2 studies. The remaining 3 studies used secondary data which was hereto unknown to the radicalization literature, namely; The
Zurich Project on the Social Development of Children and Youth survey, the Young
in Oslo survey, and the Ethnic Minority British Election Study (EMBES).
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
41
Methodologies
The average sample size for the studies was 3105.7, with the smallest sample being
from a longitudinal study with an N = 46 and the largest sample being from the
study of Berrebi (2007) comparing Palestinian terrorists to the general population,
with an N of 41,828. The combined N of the studies was 179,968, however, in
removing overlapping samples, the combined N was 172,727. Of studies examining
radical attitudes and intentions, the average sample size was 1843. For studies
examining radical behaviors, the average sample size was 5310 including the
Berrebi (2007) study and 2876 without.
The studies could be categorized as having used one of two primary methodologies; regression models and structural equation modelling (SEM). While SEM is a
type of two-stage regression model, the technique is distinct from traditional linear
models. With regards to the studies employing traditional regression models, a mix
of different approaches were taken, such as ordinary least square regressions, and
logistic regression, including traditional binomial models, as well as PROBIT models. The majority of the studies (85%) also provided bivariate estimates in the form
of correlation matrices and descriptive statistics.
Most Commonly Examined Factors
The most commonly examined factors were for the outcome of radical attitudes, for
which there were 52 factors, followed by radical behaviors, with 21 factors, and
radical intentions with 16 factors. With few exceptions, the factors identified for
radical intentions and radical behaviors were all included in the longer list of those
identified for radical attitudes. The primary difference was that radical attitudes
itself was a factor for radical intentions and behaviors as well. Table 1 includes
descriptions of the different factors identified.5
The most commonly examined factors were socio-demographic factors, with age
(N = 33) and gender (N = 32), education (N = 27), and socio-economic status
(N = 17)6 including the greatest number of effect sizes. Of all other factors, the
greatest number of effects were found for factors pertaining religious/national identity (N = 21), collective relative deprivation (N = 19), and combined, factors pertaining to law legitimacy and abidance.
Not included in this table are socio-demographic factors that do not require descriptions, such as;
age, gender, education, employment, socio-economic status, immigrant status, marital status, and
criminal history.
6
Effect sizes for socio-economic status were found only for radical attitudes.
5
42
M. Wolfowicz et al.
Table 1 Descriptions of identified factors
Factors
Adjusted personality disorder
(APD)
Aggression
Anger/hate
Example descriptions
DSM-IV personality disorders, Narcissistic Personality
Inventory
Buss-Perry Aggression Questionnaire (BPAQ-SF)
Angry, resentful, furious, or displeased with a given issue or
situation
Anxiety
General Anxiety Disorder Scale (GAD-7), Hopkins-25 scale
(HSCL-25)
Authoritarianism/
Submission to higher authority, aggression to out-groups, and
fundamentalism
adherence to traditionalism
Collective relative deprivation Feeling that in-group is deprived or discriminated against
relative to other groups
Coping skills
Ability to cope with conflict and negative encounters/situations
Depression
Patient Health Questionnaire (PHQ-9), Hopkins-25 scale
(HSCL-25),
Discrimination
Experienced personal discrimination based on identity
Experienced violence
Perpetrated or was a victim of violence involving strangers,
bullies, or parents
Group superiority
Believing that one’s in-group is better than other groups
Individual relative deprivation Feeling unfairly treated compared to others
Institutional trust
Confidence in institutions (e.g. police, parliament, courts, social
services etc.)
Integration
Attachment to/alienation from the society in which one lives
Law abidance
There is a duty to follow and abide by the law
Law legitimacy
Respect for the government/law/authorities
Legal cynicism
Believing that laws are made to be broken
Moral neutralizations
Justifications of deviant behaviors (e.g. drugs, violence etc.)
Parental involvement
Parents show interest, praise, and are aware of children’s’
whereabouts
Perceived injustice
Feeling that the individual or group is treated unjustly most of
the time
Personal strain
Loss of parents, loss of work, experienced traumatic event etc.
Personal trust
Believing that most people can be trusted
Police contact
Number of contacts respondents had with police in the previous
12-month period
Political efficacy
Having influence or being represented in the political sphere
Political grievance
Opposition to foreign intervention in the Middle East
Political participation
Participation in political party, organization, completing
surveys, petitions etc.
Political satisfaction
Satisfaction with current system of government
Radical attitudes
Support for or justification of radical violence
Realistic threat
Compared to in-group, an out-group has too much power and
influence
Religious/national identity
Strength of identity/affinity with religious/national group
Religiosity
Importance of religion in daily life and activities
(continued)
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
43
Table 1 (continued)
Factors
School bonding
Segregationist
Self-control
Similar peers
Symbolic threat
Thrill-seeking/risk-taking
Uncertainty
West Vs. Islam
Example descriptions
Enjoying going to school and/or studying, feeling an attachment
to school
Separation of people by ethnic group
Impulsivity, quick to anger
Proportion of friends of similar background
Out-groups consider themselves to be better than my in-group
Taking risks just for fun of it, without thinking of consequences,
even if dangerous
Anxiety or anticipation prior to confronting potentially harmful
events
The West is trying to attack/dominate Islam/Islamic countries
∗
These are general descriptions that we have created based on the different descriptions given in
the included studies. Factors are arranged in alphabetical order
Results from Meta-analysis
The results of a meta-analysis conducted in this study have already been published
in a technical report.7 As the below forest plots show, a wide range of pooled effect
sizes were identified for the factors associated with each of the three outcomes. With
regards to radical attitudes, the largest negative estimate was found for law abidance. For positive factors the largest estimate was highly similar peers, although
with the estimate falling below the level of statistical significance, the largest significant estimate was for authoritarian/fundamentalist personality. On the other
hand, the smallest, and sometimes statistically non-significant effects were found
for socio-demographic factors, such as socio-economic status (negative effects),
and being a second generation immigrant (positive effects) (See Figs. 7 and 8).
With regards to radical intentions, age had the largest of the negative estimates,
considerably larger than for radical attitudes, followed by out-group friendships.
For positive estimates, the largest was for correlation between radical attitudes and
radical intentions, followed by activism intentions (See Fig. 9). However, for both
radical attitudes and radical intentions, similar factors were found to have estimates
that clustered together at the upper end of the rank-order, such as symbolic and
realistic threat. Additionally, the relatively large estimates found for in-group superiority and connectedness can be juxtaposed with the negative estimate for out-­
group friendships as an indication of the relative magnitudes of these related risk
and protective factors.
Similarly, for radical behaviors, the largest negative estimates were similar to
those for radical attitudes, namely law obedience and legitimacy. With regards to
positive effects, the largest estimate was found for radical attitudes, as was the case
for the outcome of radical intentions. Additionally, similar to factors identified for
7
The technical report is available on the PROTON website https://www.projectproton.eu.
44
M. Wolfowicz et al.
Anxiety (-.015 NS)
Politcial Participate (-.017 NS)
Education (Ord.) (-.027 NS)
Marital (-.045 NS)
SES (Ord) (-.048 NS)
SES (low) (-.065 NS)
Age (-.074**)
Female (-.095***)
General trust (-.109*)
Political Satisfaction (-.119 NS)
Inst. Trust (-.144*)
2nd gen immigrant (-.148*)
Education (high) (-.153*)
Depression (-.157†)
School bonding (-.201 NS)
School performance (-.256***)
Parental Involvement (-.303***)
High integration (-.355***)
Coping skills (-.36**)
Law abidance (-1.36***)
-1.8
-1.3
-0.8
-0.3
0.2
Fig. 7 Protective factors for radical attitudes
radical attitudes, for radical behaviors factors such as thrill-seeking/risk-taking
behavior, differential associations, low self-control, criminal history, and authoritarian/ fundamentalist personality figured prominently at the upper-end of the rank-­
order of positive estimates. The key difference between radical attitudes and radical
behaviors appears to be unemployment. The estimate for unemployment on radical
behaviors was almost triple the size as the estimate for radical attitudes (See Fig. 10).
Discussion and Conclusions
According to the findings of Schuurman’s (2018) review, 7.1% of terrorism studies
employ some form of inferential statistics. This finding indicated a significant uptick
in the rigour of the quantitative literature compared to earlier reviews. Neumann and
Kleinmann (2013), found that 20% of radicalization studies quantitative and rated
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
Political efficacy (.022 NS)
Uncertainty (.033 NS)
Worship attendance (.049 NS)
West Vs Islam (.08*)
Immigrant (.084**)
Welfare recepient (.108**)
Unemployment (.116*)
Religiosity (.145*)
Discrimination (.154**)
Political Grievance (.16**)
Prayer frequency (.172***)
Violent media Exp. (.175***)
Perceived injustice (.172***)
Violence exposure (.186***)
Male (.203***)
APD/Narcissism (.213 NS)
NSM posting (.219**)
Aggression (.226**)
SES (High) (.242 NS)
Relig/Nat identity (.258***)
Personal strains (.267***)
Anti Democratic (.275*)
Ind. Rel. Dep. (.285**)
Educ. Low (.313***)
Coll. Rel. Dep. (.332***)
Anger/Hate (.34 NS)
Low integration (.376***)
Deviant peers (.416***)
Legal cynicism (.423*)
Segregation (.459***)
Moral neutralization (.462*)
Law legitimacy (.554***)
Low Self Control (.588**)
Thrill/risk seeking (.624***)
Criminal History (.678**)
Symbolic threat (.688***)
Police Contact (.721***)
Realistic threat (.761***)
Group superiority (.847***)
Authoritarian/fundamentalism (.857***)
4 3 2 1
1 2 3
1 2 3 4 5 6 7 8 9
-0. -0. -0. -0. 0 0. 0. 0. 0. 0. 0. 0. 0. 0. 1 1. 1. 1.
Fig. 8 Risk factors for radical attitudes
45
46
M. Wolfowicz et al.
Age
Outgroup friends (-.202***)
Social Connectedness (-.43**)
Perceived injustice (.213***)
Ind. Rel. Dep. (.362*)
Collect. Rel. Dep. (.427***)
Relig/National identity (.499*)
In-group connect. (.554***)
Symbolic threat (.643 †)
Realistic threat (.703**)
Ingroup superiority (.805***)
Radical beliefs (1.112***)
-1.5
-1
-0.5
0
0.5
1
1.5
Fig. 9 Risk and protective factors for radical intentions/willingness
94% of these as being of high methodological rigor. Given that their study was made
up of 260 studies, this would mean that 52 studies included in their review were
quantitative. While Neumann and Kleinmann’s review was carried out prior to the
publication of many of the studies included in our review, the similar number of
included studies is intuitive. In focussing on radicalization studies on risk factors,
the current review identified that the quantitative research on risk factors may
account for a significant proportion of the quantitative radicalization research.
Most of the studies included in this review were of a fairly high quality in terms
of their methodologies and analytic strategies. We didn’t assess quality specifically
since our inclusion criteria already included provisions relating to appropriate statistical controls and sample size. As noted above, the studies included in this review
can be divided into two primary groups of methodologies; regression models and
SEM models. It is debatable as to whether SEM type approaches are to be considered more or less robust than traditional regression models. However, we found that
SEM models were more likely to be employed in small sample studies, in which
cases such an approach may be preferable. While studies employing SEM are less
likely to include key control variables (e.g. age, gender, etc.), Pauwels and Hardyns
(2018) present arguments against the routine inclusion of such variables in radicalization research. Indeed, in their study employing generalized linear models, they
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
47
Law Obedience (-.446***)
Legitimacy (-.334***)
Age (-.325*)
School Bonding (-.213***)
Marital status (-.169†)
Parental involve. (-.112***)
Education (Ord.)(.119 NS)
Abused (.135***)
Welfare recepient (.137**)
Poor integration (.21***)
Unemployed (.3**)
Immigrant (.301 NS)
Citizen (.345 NS)
Thrill/risk seeking (.357***)
Deviant peers (.362***)
Authoritarian/fundamentalist (.374***)
Criminal hist. (.391*)
Radical beliefs (.415***)
Low Self-control (.427***)
-1
-0.5
0
0.5
1
1.5
Fig. 10 Risk and protective factors for radical behaviors
do not include these variables even though their dataset, as used in previous studies,
included them. Nevertheless, most studies using regression models included such
variables and in some studies these were the primary variables of interest, especially
for studies examining risk factors for terrorists (Perliger et al. 2016; LaFree et al.
2018; Jasko et al. 2017; Berrebi 2007; Krueger 2008; Altunbas and Thornton 2011;
Ljujic et al. 2020).
The overwhelming majority of the quantitative radicalization research that has
been produced to date has come from Europe and/or focussed on European contexts. This could in part be due to the EU’s funding of projects. Of the European
studies included in this review, at least 12 are known to have received funding from
the EU. Sources of funding have long been noted to play an important role in directing the trajectories of terrorism research (Reid 1997; Neumann and Kleinmann
2013; Sageman 2014; Silke 2007).
Overall, the research is heavily geared towards examining cognitive radicalization in the form of radical attitudes. The most common approaches to examining
48
M. Wolfowicz et al.
radical attitudes have been through the adoption of dichotomous variables examining whether or not one supports or justifies terrorism, usually with reference to
suicide bombings or specific acts of terrorism. One of the findings that stands out
most about this portion of the literature is that while a number of studies using secondary data, such as the Pew report, the overwhelming majority of studies are based
on new, original data collected by the researchers. With regards to the studies examining radical intentions, all of them were based on new, original data. In all but one
case, studies examining radical intentions included a measure of radical attitudes as
well, generally in the form of assessing support for others’ involvement in radical
violence or terrorism.
Background characteristics and demographic factors were found to be the most
commonly examined risk factors, especially in studies examining terrorists specifically. However, as has previously been pointed out, with the exception of age and
gender, factors such as employment and education are characterized by mixed and
often contradictory evidence (Victoroff 2005). Stern (2016) has previously criticized the literature’s focus on such factors and its lack of attention to the types of
psychological, personality trait and attitudinal factors traditionally examined in
criminology. However, these types of factors figured prominently among the factors
identified in this review for their relationship to radical attitudes, radical intentions,
and even self-reported radical violence.
Madriaza and Ponsot’s (2015) review emphasized the importance of examining
protective factors and found that most studies preferred to examine risk factors over
protective factors. As Lösel et al. (2018) emphasize, protective factors are not simply the opposite or reverse of risk factors. For example, being employed is not necessarily a protective factor to the same degree to which being unemployed is a risk
factor. Such factors would need to be measured separately in order to identify their
individual effects. For example, while Doosje et al. (2012, 2013) examined societal
disconnectedness as a risk factor, Van Bergen et al. (2015, 2016) examined societal
connectedness as a protective factor. Only one of the studies included in this review
were framed as analyses of protective factors specifically. Feddes et al. (2015)
examined the impact of empathy and self-esteem building as protective factors
against radical attitudes and radical intentions. However, a large number of the studies included protective factors explicitly in their analyses, or otherwise found protective effects of the independent variables which they analyzed. This included
studies examining self-reported radical violence (e.g. Pauwels and Schils 2016;
Baier et al. 2016).
The results of the meta-analysis should therefore be taken in such a context. For
example, for radical attitudes, being a female had an estimate less than half of the
size for being a male. Similarly, with regards to socio-economic status (SES), low
SES had an estimate about half the size of high SES, although neither were statistically significant. With regards to education, the estimate for lower education was
almost double the size of the estimate for higher education. Measured on an ordinal
scale, the estimate for higher education was even smaller. Also for radical attitudes,
while high law legitimacy’s estimate was quite similar in size to the risk effect for
low legitimacy, the estimate for law abidance was considerably larger than the risk
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
49
effect for legal cynicism. For radical intentions, the estimate for in-group connectedness was considerably larger than for out-group friendships, while the estimate
for employment was somewhat larger than unemployment and the estimate for high
integration was considerably larger than that of low integration.
Taken together, these results show that traditional criminogenic and criminotrophic factors have the largest risk and protective estimates respectively. Factors such
as differential associations, criminal history, and low self-control, as well as factors
relating to social control dimensions, such as parental, peer and school bonds, and
normative values, were found to have estimates with the largest magnitudes.
Overall, the results of this review indicate that the quantitative literature on risk
factors for radicalization is a burgeoning field. Employing rigorous methodologies
and increasingly producing results based off of original data, the literature is also
moving towards the analysis of more psychological and personality trait related factors. The analysis of such factors, at present, has been limited to studies examining
the two components of cognitive radicalization; attitudes and intentions.
Unfortunately, there continues to be a lack of studies comparing those with radical
attitudes to those who have exhibited radical behaviors. The majority of studies
analyzing radical behaviors, either self-reports or terrorists, use the general, non-­
radical population as a control group. The lack of studies comparing violent and
non-violent radicals represents a significant gap in the literature. Comparing these
two groups would be the ideal way to identify which risk factors may explain the
move from radical attitudes to radical behaviors. Studies with samples of confirmed
terrorists may seek to utilize survey data from non-violent radicals in order to construct more useful comparison groups. Going forward, radicalization research
should seek to further explore the relationship between radical attitudes, radical
intentions and radical behaviors, ideally through the use of more longitudinal
studies.
References
Altunbas, Y., & Thornton, J. (2011). Are homegrown Islamic terrorists different? Some UK evidence. Southern Economic Journal, 78(2), 262–272.
Baier, D., Manzoni, P., & Bergmann, M. C. (2016). Einflussfaktoren des politischen Extremismus
im Jugendalter—Rechtsextremismus, Linksextremismus und islamischer Extremismus im
Vergleich/Influencing factors of political extremism in adolescence—right-wing extremism,
left-wing extremism and Islamic extremism compared. Monatsschrift für Kriminologie und
Strafrechtsreform/Journal of Criminology an Penal Reform, 99(3), 171–198.
Beck, L., & Ajzen, I. (1991). Predicting dishonest actions using the theory of planned behavior.
Journal of Research in Personality, 25(3), 285–301.
Berrebi, C. (2007). Evidence about the link between education, poverty and terrorism among
Palestinians. Peace Economics, Peace Science and Public Policy, 13(1), 1–76.
Brockhoff, S., Krieger, T., & Meierrieks, D. (2015). Great expectations and hard times: The
(nontrivial) impact of education on domestic terrorism. Journal of Conflict Resolution, 59(7),
1186–1215.
50
M. Wolfowicz et al.
Campana, A., & Lapointe, L. (2012). The structural “root” causes of non-suicide terrorism: A
systematic scoping review. Terrorism and Political Violence, 24(1), 79–104.
Christmann, K. (2012). Preventing religious radicalisation and violent extremism: Systematic
Review of the Research Evidence. London: Youth Justice Board for England and Wales.
Dalgaard-Nielsen, A. (2018). Patterns of disengagement from violent extremism: A stocktaking
of current knowledge and implications for counterterrorism. In Expressions of Radicalization
(pp. 273–293). Cham: Palgrave Macmillan.
Doosje, B., van den Bos, K., Loseman, A., Feddes, A. R., & Mann, L. (2012). “My In-group
is Superior!”: Susceptibility for radical right-wing attitudes and behaviors in Dutch youth.
Negotiation and Conflict Management Research, 5(3), 253–268.
Doosje, B., Loseman, A., & Van Den Bos, K. (2013). Determinants of radicalization of Islamic
youth in the Netherlands: Personal uncertainty, perceived injustice, and perceived group threat.
Journal of Social Issues, 69(3), 586–604.
Ducol, B., Bouchard, M., Davies, G., Ouellet, M., & Neudecker, C. (2016). Assessment of the
state of knowledge: Connections between research on the social psychology of the Internet
and violent extremism (pp. 16–05). Waterloo: TSAS The Canadian Network for Research on
Terrorism, Security, and Society.
Enders, W., & Hoover, G. A. (2012). The nonlinear relationship between terrorism and poverty.
American Economic Review, 102(3), 267–272.
Farrell, M. M., Findley, M. G., & Young, J. (2019). Geographical approaches in the study of terrorism. In The Oxford handbook of terrorism. Oxford: Oxford University Press.
Feddes, A. R., Mann, L., & Doosje, B. (2015). Increasing self-esteem and empathy to prevent
violent radicalization: A longitudinal quantitative evaluation of a resilience training focused on
adolescents with a dual identity. Journal of Applied Social Psychology, 45(7), 400–411.
Folk, J. B., Stuewig, J. B., Blasko, B. L., Caudy, M., Martinez, A. G., Maass, S., et al. (2018). Do
demographic factors moderate how well criminal thinking predicts recidivism? International
Journal of Offender Therapy and Comparative Criminology, 62(7), 2045–2062.
Gough, D., Thomas, J., & Oliver, S. (2012). Clarifying differences between review designs and
methods. Systematic Reviews, 1(1), 28.
Hammerstrøm, K., Wade, A., Hanz, K., & Jørgensen, A. (2009). Searching for studies: Information
retrieval methods group policy brief. In Information retrieval methods group policy brief Oslo.
Norway: The Campbell Collaboration.
Hegghammer, T. (2010). The rise of Muslim foreign fighters: Islam and the globalization of Jihad.
International Security, 35(3), 53–94.
Higgins, J. P., & Green, S. (Eds.). (2011). Cochrane handbook for systematic reviews of interventions (Vol. 4). John Wiley & Sons: Hoboken
Higginson, A., Mazerolle, L., Ham Benier, K., & Bedford, L. (2014). Predictors of youth gang
membership in low-and middle-income countries: A systematic review. Campbell Systematic
Reviews, 10, 1–67.
Higginson, A., Benier, K., Shenderovich, Y., Bedford, L., Mazerolle, L., & Murray, J. (2018).
Factors associated with youth gang membership in low-and middle-income countries: A systematic review. Campbell Systematic Reviews, 14(1), 1–128.
Jasko, K., LaFree, G., & Kruglanski, A. (2017). Quest for significance and violent extremism: The
case of domestic radicalization. Political Psychology, 38(5), 815–831.
Khan, K. S., Kunz, R., Kleijnen, J., & Antes, G. (2003). Five steps to conducting a systematic
review. Journal of the Royal Society of Medicine, 96(3), 118–121.
Khosrokhavar, F. (2014). Radicalisation. Éditions de la Maison des sciences de l’homme.
Koopmans, R. (2015). Religious fundamentalism and hostility against out-groups: A comparison
of Muslims and Christians in Western Europe. Journal of Ethnic and Migration Studies, 41(1),
33–57.
Krueger, A. B. (2008). What makes a homegrown terrorist? Human capital and participation in
domestic Islamic terrorist groups in the USA. Economics Letters, 101(3), 293–296.
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
51
Kruglanski, A. W., Chen, X., Dechesne, M., Fishman, S., & Orehek, E. (2009). Yes, no, and maybe
in the world of terrorism research: Reflections on the commentaries. Political Psychology,
30(3), 401–417.
LaFree, G., & Freilich, J. (2016). Bringing criminology into the study of terrorism. The handbook
of the criminology of terrorism, 3, 1–14.
LaFree, G., Jensen, M. A., James, P. A., & Safer-Lichtenstein, A. (2018). Correlates of violent
political extremism in the United States. Criminology, 56(2), 233–268.
Laycock, G. (2012). In support of evidence-based approaches: A response to Lum and Kennedy.
Policing: A Journal of Policy and Practice, 6(4), 324–326.
Ljujic, V., Versteegt, I., Weerman, F., Thijs, F., van Prooijen, J.W., el Bouk, F., & van de Weijer,
S. (2020). Testing a Threat Model of Terrorism: A multi-method study about socio-economic
and psychological influences on terrorism involvement in the Netherlands, in Weisburd, D.,
Savona, E., Hasisi, B., Calderoni, F. (Eds.) Understanding Recruitment to Organized Crime
and Terrorism, Springer.
Liem, M., van Buuren, J., de Roy van Zuijdewijn, J., Schönberger, H., & Bakker, E. (2018).
European lone actor terrorists versus “common” homicide offenders: An empirical analysis.
Homicide Studies, 22(1), 45–69.
Lösel, F., King, S., Bender, D., & Jugl, I. (2018). Protective factors against extremism and violent radicalization: A systematic review of research. International Journal of Developmental
Science, 12(1–2), 89–102.
Lum, C., Kennedy, L. W., & Sherley, A. J. (2006). The effectiveness of counter-terrorism strategies. Campbell Systematic Reviews, 2(2), 1–50.
Lyons, H. A., & Harbinson, H. J. (1986). A comparison of political and non-political murderers in
Northern Ireland, 1974–84. Medicine, Science and the Law, 26(3), 193–198.
Macdougall, A. I., van der Veen, J., Feddes, A. R., Nickolson, L., & Doosje, B. (2018). Different
strokes for different folks: the role of psychological needs and other risk factors in early radicalisation. International Journal of Developmental Science, 12(1–2), 37–50.
Madriaza, P., & Ponsot, A. S. (2015). Preventing Radicalization: A Systematic Review. Canada:
International Centre for the Prevention of Crime.
McCauley, C. (2012). Testing theories of radicalization in polls of US Muslims. Analyses of Social
Issues and Public Policy, 12(1), 296–311.
McCauley, C., & Moskalenko, S. (2017). Understanding political radicalization: The two-­pyramids
model. American Psychologist, 72(3), 205–216.
McCauley, C., & Scheckter, S. (2008). What's special about US Muslims? The war on terrorism
as seen by Muslims in the United States, Morocco, Egypt, Pakistan, and Indonesia. Studies in
Conflict & Terrorism, 31(11), 1024–1031.
McCleary, D. F., Quillivan, C. C., Foster, L. N., & Williams, R. L. (2011). Meta-analysis of correlational relationships between perspectives of truth in religion and major psychological constructs. Psychology of Religion and Spirituality, 3(3), 163–180.
McGilloway, A., Ghosh, P., & Bhui, K. (2015). A systematic review of pathways to and processes associated with radicalization and extremism amongst Muslims in Western societies.
International Review of Psychiatry, 27(1), 39–50.
Moskalenko, S., & McCauley, C. (2009). Measuring political mobilization: The distinction
between activism and radicalism. Terrorism and political violence, 21(2), 239–260.
Mudde, C. (2004). The populist zeitgeist. Government and Opposition, 39(4), 541–563.
Munton, T., Martin, A., Lorenc, T., Marrero-Guillamon, I., Jamal, F., Lehmann, A., Sexton, M.
(2011). Understanding vulnerability and resilience in individuals to the influence of Al Qaida
violent extremism: A Rapid Evidence Assessment. Technical Report. Home Office Occasional
Paper 98, London.
Murray, J., Shenderovich, Y., Gardner, F., Mikton, C., Derzon, J. H., Liu, J., & Eisner, M. (2018).
Risk factors for antisocial behavior in low-and middle-income countries: A systematic review
of longitudinal studies. Crime and Justice, 47(1), 255–364.
Neumann, P. R. (2013). The trouble with radicalization. International Affairs, 89(4), 873–893.
52
M. Wolfowicz et al.
Neumann, P., & Kleinmann, S. (2013). How rigorous is radicalization research? Democracy and
Security, 9(4), 360–382.
Pauwels, L. J., & Hardyns, W. (2018). Endorsement for extremism, exposure to extremism via
social media and self-reported political/religious aggression. International journal of developmental science, (Preprint), 1–19.
Pauwels, L., & Schils, N. (2016). Differential online exposure to extremist content and political
violence: Testing the relative strength of social learning and competing perspectives. Terrorism
and Political Violence, 28(1), 1–29.
Perliger, A., Koehler-Derrick, G., & Pedahzur, A. (2016). The gap between participation and violence: Why we need to disaggregate terrorist ‘profiles’. International Studies Quarterly, 60(2),
220–229.
Ranstorp, M. (2006). Introduction: mapping terrorism research–challenges and priorities. In
Mapping terrorism research (pp. 15–42). New York: Routledge.
Reid, E. F. (1997). Evolution of a body of knowledge: An analysis of terrorism research.
Information Processing & Management, 33(1), 91–106.
Reid, E. F., & Chen, H. (2007). Mapping the contemporary terrorism research domain. International
Journal of Human-Computer Studies, 65(1), 42–56.
Sageman, M. (2004). Understanding terror networks. Philadelphia: University of Pennsylvania
Press.
Sageman, M. (2014). The stagnation in terrorism research. Terrorism and Political Violence, 26(4),
565–580.
Scarcella, A., Page, R., & Furtado, V. (2016). Terrorism, radicalisation, extremism, authoritarianism and fundamentalism: A systematic review of the quality and psychometric properties of
assessments. PLoS One, 11(12), e0166947.
Schmid, A. P. (2011). The literature on terrorism. In The routledge handbook of terrorism research
(pp. 475–492). New York: Routledge.
Schmid, A. P. (2013). Radicalisation, de-radicalisation, counter-radicalisation: A conceptual discussion and literature review. ICCT Research Paper, 97(1), 22.
Schmid, J., & Jongman, A. (1983). Political terrorism: A new guide to actors, authors, concepts,
data bases, theories.
Schmid, A. P., & Jongman, A. J. (1988). Political terrorism: A new guide to actors, authors, concepts, databases, theories and literature, Amsterdam. New Brunswick: SWIDOC, Transaction
Books.
Schuurman, B. (2018). Research on terrorism, 2007–2016: A review of data, methods, and authorship. Terrorism and Political Violence, 1–16.
Schuurman, B., & Eijkman, Q. (2013). Moving terrorism research forward: The crucial role of
primary sources. ICCT Background Note, 4.
Sedgwick, M. (2010). The concept of radicalization as a source of confusion. Terrorism and political violence, 22(4), 479–494.
Shenderovich, Y., Eisner, M., Mikton, C., Gardner, F., Liu, J., & Murray, J. (2016). Methods for
conducting systematic reviews of risk factors in low-and middle-income countries. BMC
Medical Research Methodology, 16(1), 32.
Silke, A. (2001). The devil you know: Continuing problems with research on terrorism. Terrorism
and political violence, 13(4), 1–14.
Silke, A. (2004). The road less travelled: Recent trends in terrorism research. In Research on terrorism: Trends, achievements and failures (pp. 186–213). London: Frank Cass.
Silke, A. (2007). The impact of 9/11 on research on terrorism. In Mapping terrorism research
(pp. 90–107). London/New York: Routledge.
Silke, A. (2009). Contemporary terrorism studies: issues in research. In Critical terrorism studies
(pp. 48–62). London: Routledge.
Smith, B. L., & Morgan, K. D. (1994). Terrorists right and left: Empirical issues in profiling
American terrorists. Studies in Conflict & Terrorism, 17(1), 39–57.
What Is the State of the Quantitative Literature on Risk Factors for Radicalization…
53
Spaaij, R., & Hamm, M. S. (2015). Key issues and research agendas in lone wolf terrorism. Studies
in Conflict & Terrorism, 38(3), 167–178.
Stankov, L., Knežević, G., Saucier, G., Radović, B., & Milovanović, B. (2018). Militant extremist
mindset and the assessment of radicalization in the general population. Journal of Individual
Differences, 39, 88–98.
Stern, J. (2016). Radicalization to extremism and mobilization to violence: What have we learned
and what can we do about it? The Annals of the American Academy of Political and Social
Science, 668(1), 102–117.
Tausch, N., Spears, R., & Christ, O. (2009). Religious and national identity as predictors of attitudes towards the 7/7 bombings among British Muslims: An analysis of UK opinion poll data.
Revue Internationale de Psychologie Sociale, 22(3), 103–126.
van Bergen, D. D., Ersanilli, E. F., Pels, T. V., & De Ruyter, D. J. (2016). Turkish-Dutch youths’
attitude toward violence for defending the in-group: What role does perceived parenting play?
Peace and Conflict: Journal of Peace Psychology, 22(2), 120–133.
Van Bergen, D. D., Feddes, A. F., Doosje, B., & Pels, T. V. (2015). Collective identity factors and
the attitude toward violence in defense of ethnicity or religion among Muslim youth of Turkish
and Moroccan Descent. International Journal of Intercultural Relations, 47, 89–100.
Van den Bos, K., Loseman, A., & Doosje, B. (2009). Waarom jongeren radicaliseren en sympathie
krijgen voor terrorisme: Onrechtvaardigheid, onzekerheid en bedreigde groepen. Research and
Documentation Centre of the Dutch Ministry of Justice.
Victoroff, J. (2005). The mind of the terrorist: A review and critique of psychological approaches.
Journal of Conflict Resolution, 49(1), 3–42.
Windisch, S., Simi, P., Ligon, G. S., & McNeel, H. (2017). Disengagement from ideologically-based
and violent organizations: A systematic review of the literature. Journal for Deradicalization,
9, 1–38.
Zhirkov, K., Verkuyten, M., & Weesie, J. (2014). Perceptions of world politics and support for
terrorism among Muslims: Evidence from Muslim countries and Western Europe. Conflict
Management and Peace Science, 31(5), 481–501.
Resilience against Political and Religious
Extremism, Radicalization, and Related
Violence: A Systematic Review of Studies
on Protective Factors
Friedrich Lösel, Doris Bender, Irina Jugl, and Sonja King
Introduction
Politically, religiously, ethnically, or otherwise motivated extremism, radicalization
towards violence, and recruitment for terrorism are urgent topics worldwide. In
spite of numerous activities to counteract these problems, there is still very limited
knowledge about valid risk factors, potential causes, developmental processes, and
effective interventions (Jensen et al. 2016; Monahan 2017; Sageman 2014; Silke
2001). This is insofar plausible as various problems complicate research and
practice:
There is no widely accepted definition of ‘terrorism’ (Schmid 2004, 2012), and
the terms ‘extremism’ or ‘radicalization’ are also used in different ways. In the present chapter, we use the following working definitions: Extremism implies a verbal
or active opposition to basic values in a society, such as democracy, equality, liberty,
rule of law, and tolerance for the faiths and beliefs of others. Radicalization is a
process by which a person adopts beliefs that justify the use of violence for social
and/or political change. Terrorism is viewed as an intentional act or threat of violence by a non-state actor that aims to attain a political, religious, economic, or
social goal, and coerces or intimidates an audience beyond the direct victims
(Doosje et al. 2016; Maskaliūnaitė 2015; Strom et al. 2017).
In empirical research, extremism and radicalization often refer to gradual differences in attitudes and behavior without clear thresholds of seriousness for violence
(Bartlett and Miller 2012; Kruglanski et al. 2009). The main target groups are
F. Lösel (*)
Institute of Psychology, University of Erlangen-Nuremberg, Erlangen, Germany
Institute of Criminology, University of Cambridge, Cambridge, UK
e-mail: fal23@cam.ac.uk; friedrich.loesel@fau.de
D. Bender · I. Jugl · S. King
Institute of Psychology, University of Erlangen-Nuremberg, Erlangen, Germany
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_3
55
56
F. Lösel et al.
h­ eterogeneous and embrace far-right, far-left, religiously motivated, nationalist/
separatist, or those who focus on single issues such as animal protection or environmental conservation (Doosje et al. 2016). However, the respective ideologies are not
necessarily distinct. For example, religious, ethnic, and nationalist motivation may
partially overlap (Doosje et al. 2016; Lösel et al. 2018) and show similarities in
developmental pathways (Ebner 2017; Jensen et al. 2016). Not all members of specific extremist groups are mainly driven by the respective ideology (e.g., Sawyer
and Hienz 2017) and underlying personality factors such as thrill-seeking may also
play a role (e.g., Pauwels and Heylen 2017).
Although plausible phases of radicalization have been proposed, the sequences
of individual needs, triggering events, ideological narratives, group affiliation, and
social identity are neither uniform nor generally lead to violent action (Borum 2011;
Jensen et al. 2016; Webber and Kruglanski 2017). Whereas most research shows
that extremist group membership plays an important core role in radicalization,
there are cases of ‘lone-actor’ terrorism that seem to be related to mental health
problems or influences from the internet (Corner and Gill 2015; Meloy and Genzman
2016). This heterogeneity questions the aim of a general theory on the topic. The
theoretical literature contains a range of explanations with concepts such as social
identity, strain, social learning, social bonding, rational choice, social movement,
group dynamics, reactance formation, intolerance of ambiguity, mental health, and
other psychosocial domains (e.g., Kruglanski et al. 2017; Raets 2017; Sageman 2017).
Methodological issues also contribute to the limited state of knowledge. Most
research is based on cross-sectional data or retrospective case analyses, whereas
sound prospective longitudinal designs are rare (Scarcella et al. 2016). Other methodological challenges are the very low base rates of the most serious cases, incomplete information in official data files, or cooperation and language barriers in
the assessments of extremists. Although there are practical approaches to counteract
terrorist plots (e.g., Perry et al. 2017; Strom et al. 2017), the low quality of many
early prevention programs of radicalization is also a methodological problem (e.g.,
Armborst and Kober 2017; Beelmann 2017).
Protective Factors and Resilience
Sound knowledge on risk factors is a key for improving approaches to prevent
extremism and radicalization. Risk factors are individual or social characteristics
that predict an enhanced probability of outcomes like the onset of extremism, radicalization towards violent behavior, or (in rare cases) terrorist acts. Risk factors can
be causes of the respective outcome, but basically, they only indicate a correlational
relationship. Kraemer et al. (2005) distinguished between merely correlative risk
factors, risk markers, and truly causal risk factors. Risk markers are variables that
may indicate an influence on the outcome, but do not have a causal impact by themselves (e.g., male sex). A causal risk factor can be assumed when the change of the
respective variable in a well-controlled intervention leads to a change in the o­ utcome
Resilience against Political and Religious Extremism, Radicalization, and Related…
57
(e.g., abstaining or desistance). Although there are many publications on disengagement and de-radicalization (e.g., Berger 2016; Raets 2017), the methodological
bases in this field do not yet allow clear causal conclusions.
The instruments for assessing risks for radicalization, extremism, and potential
terrorism are based on theoretical assumptions and correlational research. Important
instruments are, for example, VERA-2R, ERG 22+, TRAP-18, MLG, IVP, RAT,
and RADAR-ItE (see King et al. 2017; Rettenberger 2016). While the contents of
some instruments are openly accessible, other measures (particularly those used by
the police or intelligence services) are not published for plausible reasons. Most risk
assessment tools apply a structured clinical judgment approach. The respective constructs and variables of the different instruments are similar, but the concrete items
vary depending on the developmental phase and context of the target group. For
example, early developmental prediction in the community addresses different
issues than risk assessment of radicalized inmates in prisons. When applied to the
latter group, the various scales correlate significantly (King et al. 2018).
These instruments are promising, but partially due to the above mentioned methodological problems, the overall validity is still limited (Monahan 2017; Sarma
2017; Scarcella et al. 2016). Similar to other forms of violence, not single risk factors, but their accumulation is most indicative (Jensen et al. 2016). However, even
when all known terrorists have a specific pattern of (mainly static) risk factors, the
majority of people with these characteristics in the general population are not violent extremists. Accordingly, there is concern about many false positives that may
be inappropriately stigmatized (Royal College of Psychiatrists 2016; Sarma 2017).
To reduce false positives, risk assessment should refer to potential protective factors. VERA-2 contains some protective items (Pressman and Flockton 2012), but
overall, the topic requires more attention. Knowledge about protective factors is
also necessary for a better understanding and theoretical explanation of radicalization processes. Doosje et al. (2016) rightly refer to ‘shields of resilience’ that interrupt the development from extremist orientation to membership in radical groups
and to violent action. Similarly, the authors assume that protective factors contribute
to de-radicalization and disengagement.
Resilience refers to phenomena such as healthy development despite a high-risk
status, maintaining competence under specific stressors, or recuperating from severe
trauma (Lösel and Bender 2003; Rutter 2012). These phenomena correspond to the
elementary biosocial mechanisms of protection, regeneration, and repair. The processes of successful adaptation to and coping with developmental risk situations
require individual and social resources that function as protective factors. These
factors may ‘explain’ why individuals with similar risk profiles show different
behavioral outcomes. Knowledge about protective factors cannot only reduce the
rate of false positives in prediction, but help to design successful preventive
approaches.
There is much less research on protective factors against extremism and radicalization than on risk factors. This situation is similar to the broader
literature on violence of young people (Lösel and Farrington 2012). Protective factors are often misunderstood as the other ‘side of the coin’ or the opposite pole of a
quantitative risk factor. However, the analysis of protective factors and processes of
58
F. Lösel et al.
resilience requires more differentiated research methods (Lösel and Bender 2003,
2017). For example, one has to investigate curvilinear relations between quantitative variables of direct protective (promotive) factors and, in particular, study buffering effects in interaction analyses and hierarchical regressions when risk factors
are present (Loeber and Farrington 2012; Lösel and Farrington 2012). For conceptual and methodological discussions of research on protective factors and resilience,
see Luthar et al. (2000), Lösel and Bender (2003), or Masten (2016).
This article investigates factors that may strengthen resilience and have a protective function in the presence of risk factors for extremism and radicalization. We
also address factors that may help to disengage and desist from extremist violence.
We carried out a systematic review of empirical studies on these topics. Since a pilot
search showed very few relevant studies on violent behavior as outcome, we also
included research on violence-related extremist attitudes. In the following, we present the methodology of our systematic review and main results of both quantitative
and qualitative analyses. Finally, we discuss the findings from an integrative perspective and address theoretical and practical implications as well as limitations.
Method
Eligibility of Studies
We included research on far right, far left, religiously motivated, nationalist/separatist, and special issue groups (Doosje et al. 2016). Protective factors were defined as
characteristics that reduced the probability of radical violent acts, the willingness to
use radical violence, or attitudes that support or justify violence. Studies examining
broader underlying propensities (e.g., right-wing authoritarianism or social dominance) were not included. In accordance with research on resilience (e.g., Lösel and
Bender 2017) and desistance from crime (Farrall et al. 2011; Shapland et al. 2012),
we also included outcomes of de-radicalization and disengagement studies (e.g.,
Barrelle 2015; Bjørgo 2011). Both concepts are multidimensional and not generally
defined (Clubb 2015; Raets 2017), but de-radicalization primarily refers to cognitive changes, whereas disengagement implies a behavioral change (i.e. desistance
from violence or leaving a terrorist group).
There was no restriction regarding time and mode of publication of the primary
studies. ‘Unpublished’ work such as dissertations or ‘grey’ literature was included.
We focused on publications in English and German language, but did not get casual
knowledge about empirical studies in other languages. All countries of origin
regarding the local context of the research and the publication were eligible. There
was also no restriction concerning the scientific discipline of authors and research
institutions.
The main goal of the systematic review was to integrate empirical studies with
quantitative results on protective factors against extremism and radicalization. The
study design could be both cross-sectional and longitudinal. Beyond correlational
Resilience against Political and Religious Extremism, Radicalization, and Related…
59
designs, we included intervention studies targeting extremism if they provided
quantitative data on specified protective factors.
Since studies on de-radicalization and disengagement are usually case-oriented
and mostly refer to individual experiences and narratives, we also included qualitative designs in our search. The opinions on quantitative versus qualitative methods
are often polarized, but in research practice, there are meaningful and content-­
oriented combinations of both (Lösel 1985). The methodological standards of qualitative research are less elaborated and more homogeneous than those for quantitative
and, in particular, experimental designs. Popay et al. (1998), Hammarberg et al.
(2016), and other authors provided some guidelines for judging qualitative research.
Accordingly, we requested that the qualitative studies must have sufficiently valid
indicators of de-radicalization, provided descriptions of the respective method of
data gathering, contained at least ten cases, and drew integrative conclusions about
the investigated protective factors. To increase the data base, we also considered
qualitative studies examining protective factors in not yet clearly radical individuals, provided they fulfilled the other above mentioned criteria. Because of the different contents and methods of the quantitative and qualitative studies, the results of
both approaches will be reported separately. Our main focus lies on the more comprehensive quantitative research; we refer to the qualitative findings more briefly.
Literature Search
The results of meta-analyses and systematic reviews strongly depend on the search
terms used in databases. In a pilot test of search terms, we used a rather general
string, e. g., “protect∗ AND terroris∗”. This led to an unmanageable number of
mainly irrelevant results. We also tested a very narrow search string, e.g., “protective AND factor∗ AND terroris∗”. This reduced the number of usable results too
much. We thus created a few rather complex search strings including the terms
right-wing, left-wing, islamis*, salafi*, radicali*, jihad*, extremis*, and terror∗
combined WITH protect*, buffer*, resilien*, factor*, disengag*, deradical*, moderat*, reject*, desist∗.
Using these strings, we searched the following databases: Cochrane Library,
Campbell Reviews, Dissertation Abstracts, MEDLINE, PubMed, EMBASE, ERIC,
German National Library, PsycINFO, Psyndex, Science Direct, Scopus, Sociological
Abstracts, Sociological Collection, and World Cat. After the elimination of duplicates, the final output resulted in 451 titles. The subsequent stepwise screening process reduced the number to only eight relevant quantitative reports. Such a low
turnout from searches in databases is not rare in systematic reviews (e.g., Schmucker
and Lösel 2011). Therefore, we also used contacts to experts in the field and screened
relevant websites as well as the references of articles. This ‘snowball method’
revealed more than 2,000 documents. Most of them had to be excluded on the title
level, some others on the abstract level. Further details of our search are presented
in the PRISMA flow diagram in Fig. 1.
Identification
60
F. Lösel et al.
Reports identified through
database searching
(n = 451)
Reports identified through
snowballing method
(n = 2,264)
Additional
reports
(n = 7)
Reports (abstracts) screened
(n = 354)
Reports excluded
(n = 318)
Full-text reports assessed
for eligibility
(n = 36)
Reports excluded
(n = 28)
Eligibility
Screening
Reports after duplicates
removed
(n = 354)
Included
Eligible reports
(n = 8)
Eligible reports
(n = 15)
Eligible reports
(n = 1)
Included in the review
nquantitative = 17 reports
nqualitative = 7 reports
Fig. 1 Flow chart of the search process in the systematic review
In total, we retrieved 24 documents that met our eligibility criteria. Seventeen
reports provided quantitative data. Some of them contained more than one data set
or separate analyses so that we could finally include 21 quantitative analyses. Seven
studies had a qualitative design of which six contained direct data on radical individuals. In sum, 28 empirical analyses (“studies”) could be included in this review.
Coding
As Lipsey and Wilson (2001) recommended for reviews with small numbers of
primary studies, we coded the content after a discussion and consensus of two members of our research team.
Main categories of our scheme can be seen in the comprehensive table in the
Appendix. Most characteristics of the primary studies could be easily coded because
they were simple facts (e.g., country of origin, publication type, sample size, and
Resilience against Political and Religious Extremism, Radicalization, and Related…
61
age of participants). Some information was more complex, in particular with regard
to the type and construct of the respective protective factors. In these cases, we
avoided too extensive interpretations and adhered as much as possible to the categories reported by the primary study authors. In some studies, we had to be more
interpretative, for example, when Baier et al. (2016) equated islamist extremism
with hostility towards Germans. In this and other cases, the two coders discussed the
issue and reached a consensus. In principle, the studies with qualitative designs
were similarly coded as the quantitative research, but there were some differences.
For example, the findings of qualitative studies were mainly based on interpretations of the authors and not on statistical tests. As the qualitative studies addressed
processes of disengagement or de-radicalization, we also coded whether there was
more a ‘push’ or a ‘pull’ influence (Altier et al. 2014; Bjørgo 2011).
Descriptive Characteristics of the Dataset
The main characteristics of the quantitative studies are shown in the Appendix.
Eleven studies addressed religious/ethnic and five ones right-wing extremism and
radicalization. The other types were less frequent. Most of the quantitative studies
had a cross-sectional design. Four studies reported effects of an intervention with a
pre-post-design without a control group.
Ten quantitative studies were carried out in Europe, three ones in the Middle
East, two ones in North America, and one in Asia and Africa respectively. Most
samples were young people below the age of 30. Six studies included school students or other groups in education. Two studies addressed convicted radical offenders. For further details of the participants of the studies, see Appendix.
Most quantitative analyses addressed both risk and protective factors, but this
article refers to protective effects only. Twelve data analyses applied logistic regressions, four ones linear regressions, six ones LISREL or other path models, and two
ones used a pre-post group comparison. Depending on the respective statistical
analyses, the protective effects were identified by significant beta weights or path
coefficients that indicated a reduction of the extremist outcome criteria. Only
Pauwels and Svensson (2017) carried out moderator analyses on interactions of the
protective factors with a present risk. Eight studies addressed violent behavior, ten
ones violence-prone attitudes, and five ones the willingness to use violence. Two
analyses formed a composite score of violent behavior and attitudes. Many outcome
data (n = 13) were self-reports of the participants. Two studies analysed answers to
written scenarios and two studies used a special database (PIRUS) that contained
information on radicalization in the U.S. (including files of convicted extremists).
The qualitative studies investigated disengagement of individuals who had been
involved in more or less violent extremist organizations. Three studies examined
former jihadists (Syria/Iraq and Indonesia). In two studies, the participants formerly
adhered to nationalist/separatist ideologies (Northern Ireland and Spain). One study
included extremists from different organizations including violent far-right, jihad-
62
F. Lösel et al.
ist, and separatist, as well as nonviolent environmental organizations in Australia.
Further characteristics of these studies are presented in Table 2 in the results section.
Whereas these six studies dealt with self-reported disengagement, there was
another study that interviewed young at-risk individuals as well as service providers
to investigate resilience against violent extremism on the individual and community
level (Weine and Ahmed 2012). Overall, the qualitative studies did not provide as
many and as detailed data on the samples, variables, and other characteristics as the
quantitative analyses.
Results
Results of the Quantitative Studies
All quantitative studies analysed more than one potentially protective factor and
some even more than ten. For reasons of parsimony and clarity, in this article we
only present results on those variables that showed a significant protective effect. Of
course we are aware of the risk of multiple statistical tests, but in the light of some
small samples in the primary studies and the rather exploratory (not theory-testing)
character of our review, we set p < .10 as a threshold. Most significant effects were
at p < .05 or lower (see Appendix). Altogether, we found 30 different variables with
an effect that could be interpreted as being protective against extremism in the presence of risk factors. Some studies only detected one protective effect, but others up
to nine or ten (Bhui et al. 2014; Baier et al. 2016). We subsumed the variables to the
categories that Lösel and Farrington (2012) used for their classification of protective
effects against violence of young people: Individual, family, school, peer group, and
community/society factors, although there is some overlap between these categories. We will not go through all the detailed findings, but concentrate on those that
based on more than one analysis and/or were relevant for different types of extremism. Table 1 presents the condensed results from the Appendix.
At the individual level, self-control protected against far-right, far-left, and religious/ethnic extremism and radicalization (Pauwels and Svensson 2017). Adherence
to law was also a protective factor against extremist attitudes in the religious/ethnic
(mainly islamist), far-right, and far-left spectrum (Baier et al. 2016). It also reduced
violent left-wing extremist behavior (Baier et al. 2016). Acceptance of police legitimacy (Pauwels and De Waele 2014) and illness/disease (Bhui et al. 2014) also had
a protective effect in more than one analysis. Somewhat surprising was the finding
that subjective deprivation of one’s socio-economic status protected against right-­
wing extremism (Fuchs 2003).
At the family level, an appreciative, positive parenting style (Baier et al. 2016;
van Bergen et al. 2016), non-violent significant others (Jasko et al. 2016), and ownership of residential property (Asal et al. 2008) had a protective effect against different kinds of extremism/radicalization.
There were also replicated protective effects at the school level. Good school
achievement and bonding to school reduced far-right and far-left extremist attitudes
Resilience against Political and Religious Extremism, Radicalization, and Related…
63
Table 1 Main results of quantitative studies on protective factors against political and religious
extremism and radicalization
Factor
Individual factors
Self-control
Empathy
Value complexity
Anxiety about getting
incarcerated
Acceptance of police
legitimacy
Adherence to law
Political disinterest
Low importance of
religion
Intensive religious
practice
Employment
Perceived personal
discrimination
Subjective
deprivation of SESa
Dissatisfaction with
quality of life
Illness, depression
Threatening life
events
Family factors
Appreciative
parenting behavior
Ownership of
residential property
Family members not
involved in violence
Significant other not
involved in violence
Incarceration of a
family member
Membership in
specific religious
group
School factors
Higher educational
level
Type of
ideology
No. of
effects
Outcome criteria
Study
rw, lw, r/e
3
Behavior
r/e
r/e
n/s
1
1
1
Attitude
Behavior
Attitude
Pauwels and Svensson
(2017)
Feddes et al. (2015)
Liht and Savage (2013)
Cragin et al. (2015)
rw, rw
2
Behavior
rw, lw, r/e,
lw
n/s
r/e, r/e
4
Attitude, behavior
Pauwels and De Waele
(2014)
Baier et al. (2016)
1
2
Willingness
Attitude
Cragin et al. (2015)
Bhui et al. (2014)
r/e
1
Attitude and behavior
Muluk et al. (2013)
mixed
rw
1
1
Behavior
Behavior
rw
2
n/s
1
Attitude, attitude and
behavior
Attitude
Jensen et al. (2016)
Pauwels and De Waele
(2014)
Fuchs (2003)
r/e, r/e, r/e
r/e
3
1
Attitude
Attitude
Bhui et al. (2014)
Bhui et al. (2014)
lw, r/e
2
Attitude,Willingness
r/e, r/e
2
Attitude
Baier et al. (2016), van
Bergen et al. (2016)
Asal et al. (2008)
n/s
1
Willingness
Cragin et al. (2015)
mixed,
mixed
n/s
2
Behavior
Jasko et al. (2016)
1
Attitude
Cragin et al. (2015)
r/e
1
Attitude
Asal et al. (2008)
r/e
1
Attitude and behavior
Muluk et al. (2013)
Cragin et al. (2015)
(continued)
64
F. Lösel et al.
Table 1 (continued)
Factor
Good school
achievement
Bonding to school
Peer group factors
Non-violent peers
Type of
ideology
rw,
lw, rw, rw
rw, lw
n/s, mixed,
mixed
More social contacts/ r/e
social networks
Contact to foreigners rw, rw
Community/society factors
Basic attachment to r/e, r/e, r/e
society
Low social capital
r/e
Migrant of the first
r/e
generation
No. of
effects
4
Outcome criteria
Attitude, behavior
2
Attitude
3
Willingness, behavior
1
Attitude
2
Attitude, attitude and
behavior
Fuchs (2003)
3
Willingness, attitude
1
1
Attitude
Attitude
van Bergen et al.
(2015, 2016)
Bhui et al. (2014)
Bhui et al. (2014)
Study
Baier et al. (2016),
Boehnke et al. (1998)
Baier et al. (2016)
Cragin et al. (2015),
Jasko et al. (2016)
Bhui et al. (2014)
Notes. rw right-wing ideology, lw left-wing ideology, r/e religious/ethnic ideology, n/s nationalist/
separatist ideology, behavior violent extremist behavior, attitude extremist attitudes, willingness
willingness to engage in violent extremist behavior, attitude and behavior composite score of
behavior and attitudes
a
Negative evaluation of own socio-economic status
and behavior (Baier et al. 2016; Boehnke et al. 1998). The finding of Muluk et al.
(2013) on a protective effect of higher education against religious/ethnic extremism
was similar.
With regard to peer group influences, contact with non-violent peers had a protective effect against mixed types of extremism and nationalist/separatist orientations (Cragin et al. 2015; Jasko et al. 2016). More specific was the finding that
contact to foreigners came along with less right-wing extremism (Fuchs 2003).
The latter result could also be subsumed under factors on the community/society
level, where a basic attachment to society was protective against religious/ethnic
extremism (van Bergen et al. 2015, 2016). That only very few other protective factors were found in this domain was insofar plausible as all primary studies gathered
data from individuals that also reflect influences at the meso- and macro-level.
Most of the significant quantitative results in our review were found in single
studies, i.e. they based on one sample, one cultural context, and one type of extremism/radicalization respectively (see Table 1). The majority of these findings are
plausible: For example, a protective function of employment (Jensen et al. 2016),
empathy with non-Muslims (Feddes et al. 2015), becoming politically disinterested
(Cragin et al. 2015), understanding complexity in values (Liht and Savage 2013),
having non-violent family members (Cragin et al. 2015), being a first generation
migrant (Bhui et al. 2014), and showing concern about getting incarcerated or the
equally deterring experience of incarcerated family members (Cragin et al. 2015).
Resilience against Political and Religious Extremism, Radicalization, and Related…
65
Table 2 Main results of qualitative studies on disengagement and de-radicalization from extremist
violence
Factor
Individual factors
Disillusionment with group leaders or members
Disillusionment with group methods
Unmet expectations
Own identity
Age, mental and physical health
Competing loyalty
Change of mind, time to think
Costs-benefit reflections
Family factors
Good family ties
Family pressure against extremism
Own parenthood
Peer group and other social factors
Social relations
Positive social experiences, social integration
Detachment from extremist group through
imprisonment
Community and society factors
Change in political situation, legitimacy conflict
Imprisonment
Type of
ideology
Studya
Push/pull
factors
r/e, n/s, mixed
r/e, mixed
r/e
mixed
n/s, mixed
r/e
r/e, n/s
r/e
1, 2, 3, 4, 5, 6
1, 2, 3, 5, 6
3, 6
1
1, 4, 5, 6
6
2, 3, 4, 5
2, 3
Push
Push
Push
Pull
Push
Pull
Mixed
Mixed
r/e, n/s
r/e
n/s
4, 6
2, 3
4, 5
Pull
Pull
Pull
r/e, n/s, mixed
r/e, mixed
n/s
1, 2, 3, 4, 5
1, 2
4
Pull
Pull
Mixed
n/s
n/s
4, 5
4, 5
Pull
Mixed
Notes. r/e religious/ethnic ideology, n/s nationalist/separatist ideology
Studies listed in the third column: (1) Barrelle (2015); (2) Chernov Hwang (2015); (3) Chernov
Hwang et al. (2013); (4) Ferguson (2016); (5) Reinares (2011); (6) van der Heide and Huurman
(2016)
a
Other single results seemed to be somewhat contradictive, such as intensive religious practice (Muluk et al. 2013) versus low importance of religion (Bhui et al.
2014), having a wider social network versus low social capital (Bhui et al. 2014), or
being dissatisfied with own quality of life (Cragin et al. 2015). Such findings should
not be generalized across countries, topics, and samples. We will address this issue
in the discussion section.
Results of the Qualitative Studies
The main results of the qualitative studies are shown in Table 2. The most common
factor leading to disengagement from violent groups was disillusionment, in particular with group leaders and/or members or methods used by the group. Desisting
individuals also reported unmet expectations, for example, regarding their own identity as a group member (Chernov Hwang et al. 2013; van der Heide and Huurman
66
F. Lösel et al.
2016). In addition to these push factors, there were also pull factors such as positive
social relations outside the radicalized group, receiving family support, and the
experience of being treated well by authorities (Barrelle 2015; Chernov Hwang
2015). Family played not only an important role through strong ties, but also through
social pressure or own parenthood (Ferguson 2016; Reinares 2011). Some studies
reported a change of mind that followed a cost-benefit-analysis of the own radicalization (Chernov Hwang 2015; Reinares 2011). These individual change processes
seem to be triggered or promoted by contextual factors like changes in the political
situation or imprisonment that detached the individual from the extremist group and
provided “time to think”. Weak physical and psychological conditions, ageing, and
mental health problems also fostered disengagement in four studies (see Table 2).
The study that included information from service providers (Weine and Ahmed
2012) suggested a range of protective factors, e.g., family awareness of risks and
safeguards, parental monitoring and supervision, family confidants and social support, access to services and helpers, parental involvement in mosques and religious
education, focus on a future in the United States, rejection of tribalism and war,
youth civic engagement, and political dialogue. Parents’ information of law enforcement and community institutions about recruiters and associates were also seen as
steps towards desistance from radical groups.
Discussion
To our knowledge, our project is the first one that carried out a systematic review of
protective factors against extremism, radicalization, and related violence. Our search
procedure yielded a large number of potentially relevant documents, but most of
them contained theoretical, political, practical, or single case discussions. Despite
our broad eligibility criteria and the inclusion of both quantitative and qualitative
designs, we finally could only include 24 empirical studies. Most of the quantitative
studies focused on early phases of radicalization and about half of them analysed
extremist attitudes and not aggressive behavior. However, early “shields of resilience” (Doosje et al. 2016) on the attitude level are important because most terrorists
were affiliated with non-violent extremism before turning violent (Nesser 2016).
Our review revealed various protective factors on the individual, family, school,
peer-group, and community level. It is promising that most of the eligible studies
were recent, which indicates a growing research interest in this topic. In the quantitative studies, the main protective factors were self-control, a basic adherence to
law, and police legitimacy, employment, positive family ties, sound school achievement and bonding, non-deviant peers, and a basic attachment to society. The qualitative studies addressed push and pull factors in the process of disengagement from
radical violence. Here, the main protective factors were disillusionment with the
radical groups and their methods, positive family ties and social experiences outside
the radical group, self-reflection about own identity and change of mind, critical life
events such as health problems, imprisonment or own parenthood. In the following,
our findings are discussed in more detail and related to theoretical concepts.
Resilience against Political and Religious Extremism, Radicalization, and Related…
67
Discussion of the Quantitative Studies
Individual factors Table 1 shows that self-control was a protective factor in various
analyses. This is consistent to research in other fields of crime and violence (Lösel
and Bender 2017). Another significant protective factor was employment (Jensen
et al. 2016), which is consistent with protective effects against and desistance from
crime (Lösel and Farrington 2012; Sampson and Laub 2003). The protective function of anxiety about being caught is also in accordance with the broader literature
on youth violence (Lösel and Farrington 2012). Anxiety is related to enhanced
arousal and may thus protect against risky behavior and thrill-seeking (Raine 2013).
The finding of empathy for non-Muslims as a protective factor (Feddes et al. 2015)
is in line with theories and results on perspective taking as a part of social information processing in violence research (Dodge and Pettit 2003; Lösel et al. 2007). The
protective effect of value complexity (Liht and Savage 2013) points in the same
direction. The ability to switch perspective and reflect different values can counteract black-and-white thinking that is a component of hostile cognitive schemata.
However, we should also be aware of the potential “double face” of protective factors (Lösel and Bender 2003): For example, if there is only selective empathy for the
own religious group (Rahimullah et al. 2013), it may not have a protective function,
but be a risk for aggression against other groups.
At first glance, the findings on a protective effect of subjective deprivation and
dissatisfaction with the own quality of life are surprising because these factors can
also constitute a risk for extremism and violence. The respective studies addressed
attitudes towards suicide attacks in the West Bank (Cragin et al. 2015), far-right
attitudes of students in Germany (Fuchs 2003), and attitudes on religious extremism
of young adults in Belgium (Pauwels and De Waele 2014). Probably there are different underlying mechanisms in these results. For example, the far-right attitudes
in the German study were related to higher self-esteem and authoritarian attitudes
that may counteract feelings of discrimination and deprivation. This is in a­ ccordance
with feelings of superiority in right-wing Flemish groups (De Waele and Pauwels
2016). Feelings of dissatisfaction may indicate internalizing behavior problems that
go along with social withdrawal and thus can protect against affiliation with deviant
groups (Lösel and Farrington 2012).
Some individual protective factors such as acceptance of police legitimacy and
adherence to law and the society (Baier et al. 2016; van Bergen et al. 2015, 2016;
Pauwels and De Waele 2014) are related to the broader social context and will be
discussed below.
Family factors Table 1 shows various protective family factors. A positive, appreciating parenting style (Baier et al. 2016; van Bergen et al. 2016) and significant
others as well as family members who are not involved in violence (Cragin et al.
2015; Jasko et al. 2016) protected against extremism and radicalization. Family
ownership of residential property also showed a protective effect (Asal et al. 2008).
This is similar to the protective function of the family’s socio-economic status that
was found in research on youth violence (Lösel and Bender 2017).
68
F. Lösel et al.
School factors Our review also showed various protective factors in the school
domain. Good school achievement, bonding to school, and a higher educational
level had a protective effect against extremism and radicalization (Baier et al. 2016;
Boehnke et al. 1998; Muluk et al. 2013). Although our synthesis (like other studies)
subsumed school achievement under school factors, this variable also indicates
above-average intelligence that has shown a more general protective effect against
violence and crime in other studies (Ttofi et al. 2016).
Peer group factors In the domain of peer relations, there were various plausible
findings. Contact with non-deviant peers had a protective effect (Cragin et al. 2015,
Jasko et al. 2016). This is in accordance with the literature on a protective effect of
non-deviant peers against youth crime and violence (Lösel and Farrington 2012). A
larger social network seemed to be protective against extremism (Bhui et al. 2014),
which may be due to more non-extremist contacts.
Community/society factors Our review only revealed few protective factors that
can be subsumed to this level. This is insofar plausible as several individual and
micro-social factors are more or less indirectly related to the broader social context.
Van Bergen et al. (2015, 2016) showed that a basic attachment to or integration into
society protects against extremism and radicalization. This finding is in accordance
with the theory on informal social control and social bonding (Hirschi 1969). Social
bonding may also strengthen the protective function of self-control (Hirschi 2004),
but this hypothesis needs more empirical support (Lösel 2017). Bhui et al. (2014)
reported that first-generation immigrants are better protected against religious/ethnic extremism than later generations. This is in accordance with research on youth
violence (Walburg 2014) and relates micro- and macro-social processes. Whereas
first generation immigrants often seek to adapt to the values and life style of the new
country, later generations seem to have an enhanced risk of delinquency, perhaps
due to integration problems, parenting behavior, and less successful careers at
school and work (Baier et al. 2009). The protective function of a basic acceptance of
the society’s values, laws, and police legitimacy (Baier et al. 2016; van Bergen et al.
2015, 2016; Pauwels and De Waele 2014) also indicates the interplay between individual and macro-social factors. Such findings are in accordance with general theories on legitimacy in formal social control (e.g., Tyler 2006). For example, right-wing
extremists are insofar ambivalent as they oppose current policies on the one hand,
but call for law and order on the other hand. Successful coping with this ambiguity
seems to have a protective function. Similarly, value complexity can be protective in
migrants who are at risk for extremism (Liht and Savage 2013). These and other
findings resonate with theories of group identity (Tajfel and Turner 1979) and the
protective function of a dual identity against prejudice (Liht and Savage 2013).
Overall, our review of quantitative studies showed various protective factors that
were not only similar to the broader criminological research on violence, but also
consistent across different forms of extremism and radicalization. Self-control and
a basic adherence to law were relevant for right- and left-wing as well as religious/
Resilience against Political and Religious Extremism, Radicalization, and Related…
69
ethnic extremism. Non-violent significant others, positive parenting, and non-­
violent peers also had a protective function against different forms of extremism and
radicalization. Bonding to school and school achievement were only protective for
right- and left-wing extremism, but not for the religious/ethnic form. This finding
indicates different developmental origins and pathways. A basic attachment to society and religion-related factors more specifically fostered protective mechanisms
against religious/ethnic extremism. However, there seems to be no general mechanism because the cultural context was relevant for the respective influence.
Qualitative Studies
Whereas the quantitative studies mostly addressed early phases of radicalization,
the qualitative studies investigated processes of de-radicalization and disengagement after involvement in violent groups and terrorism. The few eligible studies had
been carried out in different countries and on various types of radical groups.
However, there was remarkable consistency in the findings on the main factors that
contributed to disengagement and desistance from violence. The “pull factors” that
helped individuals to desist were partially similar to the protective factors against
extremism and radicalization in the quantitative studies. Ties to or pressure from the
family, positive social relations and experiences, health problems, critical life events
like fatherhood, weighing the costs and benefits of extremist violence, and self-­
reflection on one’s own identity were relevant for disengagement (see Table 2).
These results are in accordance with the more general research on desistance from
crime and violence (Farrall et al. 2011; Sampson and Laub 2003). They are also
similar to what we found in a study on hard-core football hooligans (Lösel and
Bliesener 2006).
The distinction between push and pull factors is not clear-cut (Raets 2017), but
meaningful for practical purposes. They indicate processes of approach and avoidance motivation and may interact with each other. Disillusionment with leaders or
members of the radical group, the methods they used, and other disappointing experiences were the main push factors in our review. These findings support case studies of the process of desisting from terrorism (e.g., Horgan 2009). Push and pull
factors are also plausible from broader theoretical perspectives.
Relations to Various Theories
Due to the available primary research, our review of research on potential “shields
of resilience” could not address specific theories. However, our findings can be
subsumed under various theories of extremism and radicalization. For example, the
family and peer group factors fit to social learning theory (Atkins and Winfree
2017) and also (partially) to strain theory (Agnew 2017). Individual propensities
70
F. Lösel et al.
like self-­control and basic adherence to law and police legitimacy are in accordance with situational action theory (Wikström and Bouhana 2017). The role of
value complexity and perspective taking can be explained by theories on intergroup conflicts and prejudice (Tajfel and Turner 1979; Figueiredo et al. 2014) and
also by theories on social information processing (Dodge and Pettit 2003). The
latter is related to social learning because experiences of violence in social contexts lead to aggression-prone cognitive schemata (Lösel et al. 2007). These and
other middle-range theories are plausible, but scholars rightly plea for more theoretical integration on extremism (Pauwels and De Waele 2014) and criminology in
general (Bruinsma 2016). We agree and see some promising concepts in the abovementioned theories as well as in a General Aggression Model (Anderson and
Bushman 2002; DeWall et al. 2011) that aims to explain how personal and situational factors influence cognitions, feelings, and physiological arousal that, in turn,
affect appraisal and decision processes leading to violent or non-violent behavior.
Bandura’s (1979) integrative approach is also very useful. It explains propensities
for aggression, situational influences, and how reinforcement and self-related cognitions lead to a change or stabilisation of behavior (for an application to politically and otherwise motivated violence, see Lösel et al. 1990). Pauwels and De
Waele (2014) described an integration of constructs from control theory, procedural justice theory, general strain theory, social learning theory, and self-control
theory that can also be applied to protective factors against extremism and
radicalization.
Although these theoretical approaches are promising, the data base on protective factors is still too small to propose general explanations. More theoretical
integration is needed, but so-called “general theories” can also be misleading
(e.g., Lösel 2017). We should be realistic about what can be empirically tested in
an integrative manner. For example, some studies in our review showed that
changes in the ­political situation of a country contributed to disengagement from
terrorism (Ferguson 2016; Reinares 2011). This can be related to decision making
and legitimacy conflicts on the individual level, but the causes for the change on
the macro-level require different theoretical approaches. With regard to the individual development, we propose the concept of accumulated risk and/or protective
factors (Lösel and Bender 2003) that lead to chain reactions into and out of
violence.
Limitations of Our Review
Our review on protective factors provides some systematic knowledge on how
“shields of resilience” against radicalization and violence may function. However,
more high-quality studies on this topic are needed. Therefore, we avoid far-reaching
generalizations.
The number of eligible quantitative studies was moderate and many results could
only be derived from one single project. As in other criminological fields, more
Resilience against Political and Religious Extremism, Radicalization, and Related…
71
replications are necessary (e.g., Lösel 2018). The definitions and operationalisations
of constructs in the studies varied, what led us to mostly adhere to the descriptions
in the primary studies. Some large surveys of student samples defined extremism on
the basis of scale distributions that did not necessarily indicate a strong propensity
for radicalization. Most quantitative studies had a cross-sectional or retrospective
design and therefore cannot provide robust knowledge on developmental processes.
The methods of data collection were mainly self-reports of young people. As different data sources on youth problem behavior show only small to medium correlations (Achenbach 2006; Lösel 2002), cross-validation by other informants would be
welcome. One should also bear in mind that Table 1 and the Appendix only contain
significant results on protective factors. Most quantitative studies investigated other
variables so that a bias of multiple testing may have been possible. As it is difficult
to detect buffering protective factors in criminological research (Loeber and
Farrington 2012; Lösel and Bender 2003), we suggest to see our review more as a
preliminary than definite answer to the question of what protects against extremism
and radicalization.
The qualitative studies in our review also have limitations. We only found a
handful of eligible studies that contained more than single case analyses of desistance from politically or religiously motivated violence. The types of terrorism and
the cultural contexts were heterogeneous, which again requires caution with regard
to generalization. As in the quantitative studies, the main data sources were self-­
reports of the (previously) violent extremists. Although the primary studies found
plausible indicators of disengagement, there remains the question whether it is a
permanent desistance or only a temporary change after imprisonment and/or interventions. However, lapses and relapses are not specific for our topic, but general
issues of desistance from crime.
The qualitative findings could not be tested for statistical significance. Although
there was consistency between the different studies, a systematic comparison is
complicated by sometimes lacking details on sample characteristics, applied methods, and the validity of narratives that were gathered in the interviews.
Practical Conclusions
In spite of the above-mentioned limitations, our findings are relevant for practice.
Studies on extremism and radicalization are often based on official data, publicly
accessible information, or single case analyses. In contrast, most of the studies in
our review directly gathered data from samples of extremist individuals. This is
important for understanding developmental trajectories that are relevant for prevention. Basically, there are two different approaches to prevent extremism and
radicalization:
One group of measures aims for control and suppression, e.g., by the police,
intelligence services, and the criminal justice system, or through technical and other
forms of situational prevention. Our review contains some findings that are relevant
72
F. Lösel et al.
for these approaches. The protective effects of a basic adherence to law, acceptance
of police legitimacy, and criminal prosecution of extremist offenders underline the
importance of consequent state reactions. With regard to imprisonment, there may,
however, be both a deterrent effect and a negative influence by other extremist
inmates. Therefore, it is necessary to carry out thorough assessments and reduce the
influence of radical leaders that differ, for example, in various aspects from Salafist
followers (King et al. 2018).
The second approach to prevention contains psychosocial and educational programs for youngsters in schools, families, and neighborhoods. Other programs are
used by social services, in prisons, or in probation to support the reintegration of
extremist offenders. In this field, the studies in our review support approaches to
strengthen family ties, modify social cognitions, reflect and practice skills that help
to be assertive against extremist group influences, promote perspective taking,
reduce prejudice, and develop a law-abiding dual identity in migrants.
There are numerous developmental prevention programs against crime and
violence that more generally address these factors. The mean effect sizes are positive (Farrington et al. 2017; Lösel 2012), but there is much variation in the outcomes that is not only due to the respective program contents. Therefore,
developmental programs that showed replicated effects should be used to prevent
radicalization, but they should also be evaluated with regard to effective contents
and framing conditions. Specific programs on inter-group prejudice showed
mainly positive effects (Beelmann and Heinemann 2014), but there is also a large
variation in the outcomes. Many programs aim to achieve de-radicalization and
disengagement. Preliminary searches in our PROTON work package suggest that
there are more than 2,000 measures and programs implemented in Europe.
However, there is not yet sound evidence of their effectiveness in well-controlled
evaluations (Armborst and Kober 2017; Beelmann 2017). We are currently working on a European review of well evaluated programs. Our first impression suggests that the recent strong increase of interventions in practice is not yet
accompanied by sound evaluations of these activities. Therefore, we wish to
remind of other experiences in prevention that good intentions are no guaranty for
effectiveness (e.g., McCord 2003).
Appendix
Detailed characteristics of quantitative studies on protective factors against political
and religious extremism and radicalization
Baier et al.
(2016)
Author/
publication
date
Asal et al.
(2008)
Germany
Country
Pakistan
lw
rw
3063
9th
graders
rw
Students
Attitude
Behavior
Attitude
(forbidding child
to join jihadist
movement)
Attitude
Age
range or Type of
M (SD) extremism Outcome
46.19
r/e
Attitude
(14.62)
(permitting child
to join jihadist
movement)
4588
4697
Sample Type of
N
sample
141
Families of
martyrs in the
jihad
Binary logistic
multilevel
regression
Multilevel
linear
regression
Multilevel
linear
regression
Type of
analyses
Logistic
regression
Appreciating parenting
style
β = −.03†
Bonding to school
β = −.09∗∗∗
Good school achievement
β = −.04∗∗
School
Good school achievement
β = −.06∗∗∗
Adherence to law
β = −.33∗∗∗
Good school achievement
OR = 0.70∗
School
School
Family
School
Individual
School
Family
Family
Type of
factor
Family
Bonding to school
β = −.07∗∗∗
Membership in specific
religious groups
b = −1.79† (SE = 1.00)
Ownership of residential
property
b = 1.70∗ (SE = 0.78)
Protective factors
Ownership of residential
property
b = −2.13† (SE = 1.16)
Resilience against Political and Religious Extremism, Radicalization, and Related…
73
Bhui et al.
(2014)
Author/
publication
date
UK
Country
Adolescents
and adults
18–45
r/e
r/e
354
608
lw
Attitude
(sympathy for
terrorism)
Attitude
(condemnation of
terrorism)
Attitude
Behavior
Age
range or Type of
M (SD) extremism Outcome
3033
Sample Type of
N
sample
Binary logistic
multilevel
regression
Multilevel
linear
regression
Multinomial
logistic
regression
Type of
analyses
Low importance of
religion
RR = 0.08∗∗
Low importance of
religion
RR = 0.07∗
Peer group
More social contact/social
networks
RR = 1.52∗∗∗
Illness
RR = 8.81∗∗
Depression
RR = 1.22∗
Low social capital
RR = 0.63∗∗∗
Threatening life events
RR = 0.47†
Individual
Individual
Community/
society
Individual
Individual
Individual
Individual
Individual
Type of
factor
Individual
Adherence to law
β = −.47∗∗∗
Protective factors
Adherence to law
β = −.40∗∗∗
Adherence to law
OR = 0.60∗∗
74
F. Lösel et al.
Country
Feddes et al. The
(2015)
Netherlands
Boehnke
Germany
et al. (1998)
Cragin et al. Palestine
(2015)
Author/
publication
date
Adolescents
Adolescents
600
46
Students
590
Sample Type of
N
sample
16.93
(2.76)
18–30
12–25
r/e
n/s
rw
Path analysis
Type of
analyses
Attitude
Willingness (low
willingness to
engage in violent
protest)
Linear
regression
analysis
Attitude (Opposed MICE-­
to suicide attacks) corrected
regression
Attitude
Age
range or Type of
M (SD) extremism Outcome
Family members not
involved in violence
b = 1.44∗ (SE = 0.22)
Non-violent peers
b = 1.44∗ (SE = 0.22)
Empathy
β = −.41∗
Protective factors
Illness
RR = 0.27†
Migrant of the first
generation
RR = 0.22∗∗
Good school achievement
β = −.19∗∗
Dissatisfaction with
quality of life
b = 1.36† (SE = 0.70)
Anxiety about getting
incarcerated
b = 0.66∗∗ (SE = 0.20)
Incarceration of a family
member
b = 0.96∗∗ (SE = 0.23)
Political disinterest
b = 0.94† (SE = 0.50)
Individual
Peer group
Family
Individual
Family
Individual
Individual
School
Community/
society
Type of
factor
Individual
Resilience against Political and Religious Extremism, Radicalization, and Related…
75
1473
81
USA
UK
Jensen et al.
(2016)
Liht and
Savage
(2013)
1496
USA
Country
Germany
Adolescents
Extremists
Extremists
Sample Type of
N
sample
4367
Students
Jasko et al.
(2016)
Author/
publication
date
Fuchs
(2003)
19.48
(2.14)
34.18
(13.22)
34.24
(13.30)
r/e
mixed
Mixed
Behavior
Behavior
Behavior
Attitude and
behavior
Age
range or Type of
M (SD) extremism Outcome
14–21
rw
Attitude
Multivariate
logistic
regression
model
Pre/post
comparison
Hierarchical
logistic
regression
Type of
analyses
Path analysis
Value complexity
25–50%∗
8.3–22.7%∗
Protective factors
Contact to foreigners
β = −.24∗∗∗
Subjective deprivation of
SES
β = −.08∗∗
Contact to foreigners
β = −.04∗∗
Subjective deprivation of
SES
β = −.05∗∗∗
Significant other not
involved in violence
b = −0.57† (SE = 0.33)
Significant other not
involved in violence
b = −0.57† (SE = 0.29)
Non-violent peers
b = −0.76∗∗ (SE = 0.24)
Non-violent peers
b =− 0.82∗∗∗ (SE = 0.17)
Employment
b = −.56∗∗ (SE = .21)
Individual
Individual
Peer group
Peer group
Family
Family
Individual
Peer group
Individual
Type of
factor
Peer group
76
F. Lösel et al.
van Bergen
et al. (2015)
The
166
Netherlands
Students
15.70
(1.02)
rw
r/e
r/e
16–24
rw
4330
Adolescents
<18 –
>22
lw
4337
Pauwels and Belgium
Svensson
(2017)
Adolescents
Type of
analyses
Path analysis
Willingness
Behavior
Behavior
Behavior
Behavior (violence
towards persons)
Multivariate
logistic
regression
Multivariate
logistic
regression
Multivariate
logistic
regression
SEM
Behavior (violence Blockwise
towards property) multiple
logistic
regression
analysis
Age
range or Type of
M (SD) extremism Outcome
39.90
r/e
Attitude and
(13.40)
behavior
4326
2879
Country
Indonesia
Sample Type of
N
sample
934
Adults
Pauwels and Belgium
De Waele
(2014)
Author/
publication
date
Muluk et al.
(2013)
Community/
society
Individual
Self-control
β = −.32∗∗∗
Basic attachment to
society
β = −.17∗
Individual
Individual
Individual
Individual
Individual
School
Type of
factor
Individual
Self-control
β = −.38∗∗∗
Perceived personal
discrimination
OR = 0.72∗
Acceptance of police
legitimacy
OR = 0.78∗
Self-control
β = −.32∗∗
Protective factors
Intensive religious practice
β = −0.15
Higher educational level
β = −0.14
Acceptance of police
legitimacy
OR = 0.71∗∗
Resilience against Political and Religious Extremism, Radicalization, and Related…
77
The
133
Netherlands
van Bergen
et al. (2016)
Students
15.58
(0.95)
r/e
SEM
SEM
Attitude
Type of
analyses
SEM
Willingness
Age
range or Type of
M (SD) extremism Outcome
15.86
r/e
Willingness
(1.02)
Protective factors
Basic attachment to
society
β = −.05∗
Appreciating parenting
style
β = −.11∗∗
Basic attachment to
society
β = −.17†
Community/
society
Family
Type of
factor
Notes. rw right-wing ideology, lw left-wing ideology, r/e religious/ethnic ideology, n/s nationalist/separatist ideology, SEM structural equation model; if the
authors applied several models in a regression analysis, the reported results always reflect the model in which all variables were included.
†
p < .10, ∗ p < .05, ∗∗ p < .01, ∗∗∗ p < .001
Country
Sample Type of
N
sample
232
Author/
publication
date
78
F. Lösel et al.
Resilience against Political and Religious Extremism, Radicalization, and Related…
79
References
∗ Quantitative studies included in this review
† Qualitative studies included in this review
Achenbach, T. M. (2006). As others see us: Clinical and research implications of cross-informant
correlations for psychopathology. Current Directions in Psychological Science, 15, 94–98.
https://doi.org/10.1111/j.0963-7214.2006.00414.x.
Agnew, R. (2017). General strain theory and terrorism. In G. LaFree & J. D. Freilich (Eds.), The
handbook of the criminology of terrorism (pp. 121–132). Chichester: Wiley-Blackwell.
Altier, M. B., Thoroughgood, C. N., & Horgan, J. G. (2014). Turning away from terrorism. Journal
of Peace Research, 51, 647–661. https://doi.org/10.1177/0022343314535946.
Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual Review of Psychology, 53,
27–51. https://doi.org/10.1146/annurev.psych.53.100901.135231.
Armborst, A., & Kober, M. (2017). Effekte von Ansätzen zur Prävention Islamistischer
Radikalisierung: Systematische Übersichtsarbeit zu den Methoden und Ergebnissen von
Studien zur Evaluation von Präventionsansätzen im Bereich Islamismus. Berichte des
Nationalen Zentrums für Kriminalprävention, Nr. 1. [The effectiveness of approaches to prevent islamist radicalization: A systematic review of the methods and results of evaluation studies. Reports of the National Centre of Crime Prevention, No. 1.] Bonn: NZK. Retrieved from
https://www.nzkrim.de/fileadmin/nzk/NZKBerichte/NZK201700101.pdf.
∗Asal, V., Fair, C. C., & Shellman, S. (2008). Consenting to a child's decision to join a Jihad:
Insights from a survey of militant families in Pakistan. Studies in Conflict and Terrorism, 31,
973–994. https://doi.org/10.1080/10576100802400201.
Atkins, J. K., & Winfree, L. T. (2017). Social learning theory and becoming a terrorist: New challenges for a general theory. In G. LaFree & J. D. Freilich (Eds.), The handbook of the criminology of terrorism (pp. 1133–1149). Chichester: Wiley-Blackwell.
Baier, D., Pfeiffer, C., Simonson, J., & Rabold, S. (2009). Jugendliche in Deutschland als Opfer
und Täter von Gewalt [Youth in Germany as victims and perpetrators of violence]. Hannover:
Kriminologisches Forschungsinstitut Niedersachsen. Retrieved from https://kfn.de/wp-content/uploads/Forschungsberichte/FB 107.pdf.
∗Baier, D., Manzoni, P., & Bergmann, M. C. (2016). Einflussfaktoren des politischen Extremismus
im Jugendalter: Rechtsextremismus, Linksextremismus und islamischer Extremismus im
Vergleich [Factors influencing adolescents’ political extremism: Right-wing, left-wing, and
Islamist extremism compared]. Monatsschrift für Kriminologie und Strafrechtsreform [Journal
of Criminology and Penal Reform], 99, 171–198. https://doi.org/10.1515/mkr-2016-0302.
Bandura, A. (1979). Sozial-kognitive Lerntheorie [Social-cognitive learning theory]. Stuttgart:
Klett-Cotta.
Barrelle, K. (2015). Pro-integration: Disengagement from and life after extremism. Behavioral
Sciences of Terrorism and Political Aggression, 7, 129–142. https://doi.org/10.1080/194344
72.2014.988165.
Bartlett, J., & Miller, C. (2012). The edge of violence: Towards telling the difference between
violent and non-violent radicalization. Terrorism and Political Violence, 24, 1–21. https://doi.
org/10.1080/09546553.2011.594923.
Beelmann, A. (2017). Grundlagen einer entwicklungsorientierten Prävention des
Rechtsextremismus. Jena: Friedrich-Schiller-Universität.
Beelmann, A., & Heinemann, K. (2014). Preventing prejudice and improving intergroup attitudes.
A meta-anaylsis of child and adolescent training programs. Journal of Applied Developmental
Psychology, 35, 10–24. https://doi.org/10.1016/j.appdev.2013.11.002.
Berger, J. M. (2016). Promoting disengagement from violent extremism. Terrorism and CounterTerrorism Studies. The Hague: International Centre for Counter-Terrorism. https://doi.
org/10.19165/2016.2.05.
80
F. Lösel et al.
∗Bhui, K., Everitt, B., Jones, E., & Correa-Velez, I. (2014). Might depression, psychosocial adversity, and limited social assets explain vulnerability to and resistance against violent radicalisation? PLoS One, 9, e105918. https://doi.org/10.1371/journal.pone.0105918.
Bjørgo, T. (2011). Dreams and disillusionment: Engagement in and disengagement from militant extremist groups. Crime, Law and Social Change, 55, 277–285. https://doi.org/10.1007/
s10611-011-9282-9.
∗Boehnke, K., Hagan, J., & Merkens, H. (1998). Right-wing extremism among German adolescents: Risk factors and protective factors. Applied Psychology, 47, 109–126. https://doi.
org/10.1111/j.1464-0597.1998.tb00016.x.
Borum, R. (2011). Radicalization into violent extremism II: A review of conceptual
models and empirical research. Journal of Strategic Security, 4, 37–62. https://doi.
org/10.5038/1944-0472.4.4.2.
Bruinsma, G. J. N. (2016). Proliferation of crime causation theories in an era of fragmentation:
Reflections on the current state of criminological theory. European Journal of Criminology,
13, 659–666.
Chernov Hwang, J. (2015). The disengagement of Indonesian jihadists: Understanding the pathways. Terrorism and Political Violence, 29, 277–295. https://doi.org/10.1080/09546553.2015
.1034855.
†Chernov Hwang, J., Panggabean, R., & Fauzi, I. A. (2013). The disengagement of Jihadis in Poso,
Indonesia. Asian Survey, 53, 754–777. https://doi.org/10.1525/as.2013.53.4.754.
Clubb, G. (2015). De-radicalization, disengagement and the attitudes-behavior debate. In
C. Kennedy-Pipe, G. Clubb, & S. Mabon (Eds.), Terrorism and political violence (pp. 258–
266). London: Sage.
Corner, E., & Gill, P. (2015). A false dichotomy? Mental illness and lone-actor terrorism. Law and
Human Behavior, 39, 23–34. https://doi.org/10.1037/lhb0000102.
∗Cragin, R. K., Bradley, M. A., Robinson, E., & Steinberg, P. S. (2015). What factors cause
youth to reject violent extremism? Results of an exploratory analysis in the West Bank. RAND
Corporation. Retrieved from http://www.rand.org/content/dam/rand/pubs/research_reports/
RR1100/RR1118/RAND_RR1118.pdf.
De Waele, M., & Pauwels, L. (2016). Why do Flemish youth participate in right-wing disruptive
groups? In C. Maxson & F. Esbensen (Eds.), Gang transitions and transformations in an international context (pp. 173–200). New York: Springer.
DeWall, C. N., Anderson, C. A., & Bushman, B. J. (2011). The general aggression model: Theoretical
extensions to violence. Psychology of Violence, 1, 245–258. https://doi.org/10.1037/a0023842.
Dodge, K. A., & Pettit, G. S. (2003). A biopsychosocial model of the development of chronic
conduct problems in adolescence. Developmental Psychology, 39, 349–371. https://doi.
org/10.1037/0012-1649.39.2.349.
Doosje, B., Moghaddam, F. M., Kruglanski, A. W., de Wolf, A., Mann, L., & Feddes, A. R. (2016).
Terrorism, radicalization and de-radicalization. Current Opinion in Psychology, 11, 79–84.
https://doi.org/10.1016/j.copsyc.2016.06.008.
Ebner, J. (2017). The rage: The vicious circle of Islamist and far right extremism. London: Tauris.
Farrall, S., Hough, M., Maruna, S., & Sparks, R. (Eds.). (2011). Escape routes: Contemporary
perspectives on life after punishment. Milton Park, UK: Routledge.
Farrington, D. P., Gaffney, H., Lösel, F., & Ttofi, M. M. (2017). Systematic reviews of the effectiveness of developmental prevention programs in reducing delinquency, aggression, and bullying. Aggression and Violent Behavior, 33, 91–106. https://doi.org/10.1016/j.avb.2016.11.003.
∗Feddes, A. R., Mann, L., & Doosje, B. (2015). Increasing self-esteem and empathy to prevent
violent radicalization: A longitudinal quantitative evaluation of a resilience training focused on
adolescents with a dual identity. Journal of Applied Social Psychology, 45, 400–411. https://
doi.org/10.1111/jasp.12307.
Ferguson, N. (2016). Disengaging from terrorism: A Northern Irish experience. Journal for
Deradicalization, 6, 1–23. Retrieved from http://journals.sfu.ca/jd/index.php/jd/article/
view/41/36.
Resilience against Political and Religious Extremism, Radicalization, and Related…
81
Figueiredo, A., Valentim, J. P., & Doosje, B. (2014). Theories on intergroup relations and emotions:
A theoretical overview. Psychologica, 2, 7–33. https://doi.org/10.14195/1647-8606_57-2_1.
∗Fuchs, M. (2003). Rechtsextremismus von Jugendlichen. Zur Erklärungskraft verschiedener
theoretischer Konzepte [right-wing extremism among youth: On the explanatory power
of various theoretical concepts]. Kölner Zeitschrift für Soziologie und Sozialpsychologie
[Cologne Journal of Sociology and Social Psychology], 55, 654–678. https://doi.org/10.1007/
s11577-003-0116-3.
Hammarberg, K., Kirkman, M., & de Lacey, S. (2016). Qualitative research methods: When to
use them and how to judge them. Human Reproduction, 31, 498–501. https://doi.org/10.1093/
humrep/dev334.
Hirschi, T. (1969). Causes of delinquency. Berkeley: University of California Press.
Hirschi, T. (2004). Self-control and crime. In R. F. Baumeister & K. D. Vohs (Eds.), Handbook of
self-regulation: Research, theory, and applications (pp. 537–552). New York: Guilford.
Horgan, J. (2009). Individual disengagement: A psychological analysis. In J. Horgan & T. Bjørgo
(Eds.), Cass series on political violence. Leaving terrorism behind. Individual and collective
disengagement (pp. 17–29). Oxon: Routledge.
∗Jasko, K., LaFree, G., & Kruglanski, A. (2016). Quest for significance and violent extremism:
The case of domestic radicalization. Political Psychology, 1, 195. https://doi.org/10.1111/
pops.12376.
∗Jensen, M., LaFree, G., James, P. A., Atwell-Seate, A., Pisoiu, D., Stevenson, J., & Picarelli, J.
(2016). Final report: Empirical assessment of domestic radicalization (EADR). Report to the
National Institute of Justice, Office of Justice Programs. Washington, DC: U.S. Department of
Justice.
King, S., Bender, D., & Lösel, F. (2017). Instrumente zur Risikobeurteilung extremistisch motivierter Straftäter: Eine Synopse [Instruments for risk assessment of extremist offenders: A synopsis]. In K. Boers & M. Schaerff (Eds.), Kriminologische Welt in Bewegung [Criminology in
motion] (pp. 191–203). Mönchengladbach: Forum Verlag.
King, S., Endres, J., Schwaß, M., Stemmler, M., Lauchs, L., & Armborst, A. (2018). Prisoners
with Islamist relations: Are prisoner files a valuable data source for individual assessment
and for research? International Journal of Developmental Science, 12, 129–141. https://doi.
org/10.3233/DEV-170243.
Kraemer, H. C., Lowe, K. K., & Kupfer, D. J. (2005). To your health: How to understand what
research tells us about risk. Oxford: Oxford University Press.
Kruglanski, A. W., Chen, X., Dechesne, M., Fishman, S., & Orehek, E. (2009). Fully committed:
Suicide bombers motivation and the quest for personal significance. Political Psychology, 30,
331–357. https://doi.org/10.1111/j.1467-9221.2009.00698.x.
Kruglanski, A., Jasko, K., Chernikova, M., Dugas, M., & Webber, D. (2017). To the fringe and
back: Violent extremism and the psychology of deviance. American Psychologist, 72, 217–230.
https://doi.org/10.1037/amp0000091.
∗Liht, J., & Savage, S. (2013). Preventing violent extremism through value complexity: Being Muslim Being British. Journal of Strategic Security, 6, 44–66. https://doi.
org/10.5038/1944-0472.6.4.3.
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Thousand Oaks: Sage Publications.
Loeber, R., & Farrington, D. P. (2012). Advancing knowledge about direct protective factors that
may reduce youth violence. American Journal of Preventive Medicine, 43, 24–27. https://doi.
org/10.1016/j.amepre.2012.04.031.
Lösel, F. (1985). Zur Kontroverse um eine gegenstandsangemessene psychologische Forschung.
Bemerkungen aus der Sicht der Forschungspraxis [On the controversy between quantitative and
qualitative methodology: Comments from the view of research practice]. In W.-F. Kugemann,
S. Preiser, & K. A. Schneewind (Eds.), Psychologie und komplexe Lebenswirklichkeit
[Psychology and complex reality] (pp. 65–91). Göttingen: Hogrefe.
Lösel, F. (2002). Risk/need assessment and prevention of antisocial development in young people:
Basic issues from a perspective of cautionary optimism. In R. Corrado, R. Roesch, S. D. Hart,
82
F. Lösel et al.
& J. Gierowski (Eds.), Multiproblem violent youth: A foundation for comparative research
needs, interventions and outcomes (pp. 35–57). Amsterdam: IOS.
Lösel, F. (2012). Entwicklungsbezogene Prävention von Gewalt und Kriminalität [Developmental
approach to the prevention of violence and delinquency]. Forensische Psychiatrie, Psychologie,
Kriminologie [Forensic Psychiatry, Psychology, Criminology], 6, 71–84. https://doi.
org/10.1007/s11757-012-0159-2.
Lösel, F. (2017). Self-control as a theory of crime: A brief stocktaking after 27/42 years. In
C. Bijlefeld & P. van der Laan (Eds.), Liber amoricum for Gerben Bruinsma (pp. 232–238).
The Hague, NL: Boom.
Lösel, F. (2018). Evidence comes by replication, but needs differentiation: the reproducibility
issue in science and its relevance for criminology. Journal of Experimental Criminology, 14,
257–278.
Lösel, F., & Bender, D. (2003). Protective factors and resilience. In D. P. Farrington & J. Coid
(Eds.), Early prevention of adult antisocial behaviour (pp. 130–204). Cambridge, UK:
Cambridge University Press.
Lösel, F., & Bender, D. (2017). Protective factors against crime and violence in adolescence. In
P. Sturmey (Ed.), The Wiley handbook of violence and aggression. New York: Wiley.
Lösel, F., & Bliesener, T. (2006). Hooliganismus in Deutschland: Verbreitung, Ursachen
und Prävention [Football hooliganism in Germany: Prevalence, origins, and prevention].
Monatsschrift für Kriminologie und Strafrechtsreform [Journal of Criminology and Penal
Reform], 89, 229–245.
Lösel, F., Bliesener, T., & Bender, D. (2007). Social information processing, experiences of aggression in social contexts, and aggressive behavior in adolescents. Criminal Justice and Behavior,
34, 330–347. https://doi.org/10.1177/0093854806295833.
Lösel, F., & Farrington, D. P. (2012). Direct protective and buffering protective factors in the
development of youth violence. American Journal of Preventive Medicine, 43, 8–23. https://
doi.org/10.1016/j.amepre.2012.04.029.
Lösel, F., King, S., Bender, D., & Jugl, I. (2018). Protective factors against extremism and violent radicalization: A systematic review of research. International Journal of Developmental
Science, 12, 89–102. https://doi.org/10.3233/DEV-170241.
Lösel, F., Selg, H., Müller-Luckmann, E., & Schneider, U. (1990). Ursachen, Prävention und
Kontrolle von Gewalt aus psychologischer Sicht [Origins, prevention, and control of violence:
The psychological perspective]. In H.-D. Schwind, J. Baumann, F. Lösel, R. E. Remschmidt,
H.-J. Kerner, et al. (Eds.), Ursachen, Prävention und Kontrolle von Gewalt. Analysen und
Vorschläge der Unabhängigen Regierungskommission zur Verhinderung und Bekämpfung von
Gewalt, Bd. 2. [Report of the German Federal Government’s Commission on origins, prevention, and control of violence] (pp. 1–156). Berlin: Duncker & Humblot.
Luthar, S. S., Cicchetti, D., & Becker, B. (2000). The construct of resilience: A critical evaluation and guidelines for future work. Child Development, 71, 543–562. https://doi.
org/10.1111/1467-8624.00164.
Maskaliūnaitė, A. (2015). Exploring the theories of radicalization. International Studies.
Interdisciplinary Political and Cultural Journal, 17, 9–26. https://doi.org/10.1515/
ipcj-2015-0002.
Masten, A. S. (2016). Resilience in developing systems: The promise of integrated approaches.
European Journal of Developmental Psychology, 13, 297–312. https://doi.org/10.1080/17405
629.2016.1147344.
McCord, J. (2003). Cures that harm: Unanticipated outcomes of crime prevention programs. The
Annals of the American Academy of Political and Social Science, 587, 16–30. https://doi.
org/10.1177/0002716202250781.
Meloy, J. R., & Genzman, J. (2016). The clinical threat assessment of the lone-actor terrorist. The
Psychiatric Clinics of North America, 39, 649–662. https://doi.org/10.1016/j.psc.2016.07.004.
Monahan, J. (2017). The individual risk assessment of terrorism. In G. LaFree & J. D. Freilich (Eds.),
The handbook of the criminology of terrorism (pp. 520–534). Chichester: Wiley-Blackwell.
Resilience against Political and Religious Extremism, Radicalization, and Related…
83
∗Muluk, H., Sumaktoyo, N. G., & Ruth, D. M. (2013). Jihad as justification: National survey
evidence of belief in violent jihad as a mediating factor for sacred violence among Muslims
in Indonesia. Asian Journal of Social Psychology, 16, 101–111. https://doi.org/10.1111/
ajsp.12002.
Nesser, P. (2016). Islamist terrorism in Europe: A history. New York: Oxford University Press.
∗Pauwels, L. J., & De Waele, M. (2014). Youth involvement in politically motivated violence: Why
do social integration, perceived legitimacy, and perceived discrimination matter? International
Journal of Conflict and Violence, 8, 134–153. https://doi.org/10.4119/UNIBI/ijcv.379.
Pauwels, L. J., & Heylen, B. (2017). Perceived group threat, perceived injustice, and self-reported
right-wing violence: An integrative approach to the explanation right-wing violence.Journal
of Interpersonal Violence, 0886260517713711. https://doi.org/10.1177/0886260517713711.
∗Pauwels, L. J. R., & Svensson, R. (2017). How robust is the moderating effect of extremist beliefs
on the relationship between self-control and violent extremism? Crime and Delinquency, 63,
1000–1016. https://doi.org/10.1177/0011128716687757.
Perry, S., Weisburd, D., & Hasisi, B. (2017). The ten commandments for effective counterterrorism. In G. LaFree & J. D. Freilich (Eds.), Wiley handbooks in criminology and criminal justice.
The handbook of the criminology of terrorism (pp. 482–494). Wiley-Blackwell: Chichester,
UK.
Popay, J., Rogers, A., & Williams, G. (1998). Rationale and standards for the systematic review
of qualitative literature in health services research. Qualitative Health Research, 8, 341–351.
https://doi.org/10.1177/104973239800800305.
Pressman, D. E., & Flockton, J. (2012). Calibrating risk for violent political extremists and terrorists: The VERA 2 structured assessment. The British Journal of Forensic Practice, 14, 237–
251. https://doi.org/10.1108/14636641211283057.
Raets, S. (2017). The we in me: Consiering terrorist desistance from a social identity perspective.
Journal for Deradicalization, 2017/18, 1–27. Retrieved from http://journals.sfu.ca/jd/index.
php/jd/article/view/122/100.
Rahimullah, R. H., Larmar, S., & Abdalla, M. (2013). Understanding violent radicalization
amongst Muslims: A review of the literature. Journal of Psychology and Behavioral Science, 1,
19–35. Retreived from https://search.ror.unisa.edu.au/?redirect=http://researchoutputs.unisa.
edu.au/11541.2%2f122231.
Raine, A. (2013). The anatomy of violence: The biological roots of crime. New York: Pantheon
Books.
Reinares, F. (2011). Exit from terrorism: A qualitative empirical study on disengagement and
deradicalization among members of ETA. Terrorism and Political Violence, 23, 780–803.
­https://doi.org/10.1080/09546553.2011.613307.
Rettenberger, M. (2016). Die Einschätzung der Gefährlichkeit bei extremistischer Gewalt und
Terrorismus. [Risk assessment in cases of extremist violence and terrorism]. Kriminalistik
[Criminalistics], 8–9, 532–537.
Royal College of Psychiatrists. (2016). Counter-terrorism and psychiatry. London: Royal College
of Psychiatrists.
Rutter, M. (2012). Resilience as a dynamic concept. Development and Psychopathology, 24, 335–
344. https://doi.org/10.1017/S0954579412000028.
Sageman, M. (2014). The stagnation in terrorism research. Terrorism and Political Violence, 26,
565–580. https://doi.org/10.1080/09546553.2014.895649.
Sageman, M. (2017). Turning to political violence: The emergence of terrorism. Philadelphia:
University of Pennsylvania Press.
Sampson, R. J., & Laub, J. H. (2003). Shared beginnings, divergent lives: Delinquent boys to age
70. Cambridge, MA: Harvard University Press.
Sarma, K. M. (2017). Risk assessment and the prevention of radicalization from nonviolence into
terrorism. American Psychologist, 72, 278–288. https://doi.org/10.1037/amp0000121.
Sawyer, J. P., & Hienz, J. (2017). What makes them do it? Individual-level indicators of extremist
outcomes. In G. LaFree & J. D. Freilich (Eds.), Wiley handbooks in criminology and criminal
justice. The handbook of the criminology of terrorism (pp. 47–61). Chichester: Wiley-Blackwell.
84
F. Lösel et al.
Scarcella, A., Page, R., & Furtado, V. (2016). Terrorism, radicalisation, extremism, authoritarianism and fundamentalism: A systematic review of the quality and psychometric properties of
assessments. PLoS ONE, 11, e0166947. https://doi.org/10.1371/journal.pone.0166947.
Schmid, A. (2004). Terrorism – the definitional problem. Case Western Reserve Journal of
International Law, 36, 375–420.
Schmid, A. (2012). The revised academic consensus definition of terrorism. Perspectives on
Terrorism, 6, 158–159. Retrieved from https://www.terrorismanalysts.com/pt/index.php/pot/
article/view/193.
Schmucker, M., & Lösel, F. (2011). Meta-analysis as a method of systematic reviews. In D. Gadd,
S. Karstedt, & S. F. Messner (Eds.), The sage handbook of criminological research methods
(pp. 425–443). Thousand Oaks: Sage.
Shapland, J., Bottoms, A., & Grant, M. (2012). Perception of the criminal justice system among
young adult would-be desisters. In F. Lösel, A. E. Bottoms, & D. P. Farrington (Eds.), Young
adult offenders: Lost in transition? (pp. 128–145). Milton Park: Routledge.
Silke, A. (2001). The devil you know: Continuing problems with research on terrorism. Terrorism
and Political Violence, 13, 1–14. https://doi.org/10.1080/09546550109609697.
Strom, K. J., Hollywood, J. S., & Pope, M. W. (2017). Terrorist plots the United States: What we
have really faced, and how we might best defend against it. In G. LaFree & J. D. Freilich (Eds.),
Wiley handbooks in criminology and criminal justice. The handbook of the criminology of terrorism (pp. 468–481). Wiley-Blackwell: Chichester.
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin
& S. Worchel (Eds.), The social psychology of intergroup relations (pp. 33–48). Monterey:
Brooks/Cole.
Ttofi, M. M., Farrington, D. P., Piquero, A. R., Lösel, F., DeLisi, M., & Murray, J. (2016).
Intelligence as a protective factor against offending: A meta-analytic review of prospective longitudinal studies. Journal of Criminal Justice, 45, 4–18. https://doi.org/10.1016/j.
jcrimjus.2016.02.003.
Tyler, T. R. (2006). Why people obey the law. Princeton: Princeton University Press.
∗van Bergen, D. D., Ersanilli, E. F., Pels, T. V. M., & de Ruyter, D. J. (2016). Turkish-Dutch
youths’ attitude toward violence for defending the in-group: What role does perceived parenting play? Peace and Conflict: Journal of Peace Psychology, 22, 120–133. https://doi.
org/10.1037/pac0000173.
van der Heide, L., & Huurman, R. (2016). Suburban bliss or disillusionment: Why do terrorists
quit? Journal for Deradicalization, 8, 1–24. Retrieved from http://journals.sfu.ca/jd/index.php/
jd/article/view/64.
∗van Bergen, D. D., Feddes, A. F., Doosje, B., & Pels, T. V. (2015). Collective identity factors and
the attitude toward violence in defense of ethnicity or religion among Muslim youth of Turkish
and Moroccan Descent. International Journal of Intercultural Relations, 47, 89–100. https://
doi.org/10.1016/j.ijintrel.2015.03.026
Walburg, C. (2014). Migration und Jugenddelinquenz: Mythen und Zusammenhänge. Ein
Gutachten im Auftrag des Mediendienst. [Migration and youth delinquency: Myths and correlations. An expert statement prepared for the media service.] Retrieved from Institut für
Kriminalwissenschaften, Universität Münster website: https://mediendienst-integration.de/fileadmin/Dateien/Gutachten_Kriminalitaet_Migration_Walburg.pdf
Webber, D., & Kruglanski, A. W. (2017). Psychological factors in radicalization: A "3N" approach.
In G. LaFree & J. D. Freilich (Eds.), Wiley handbooks in criminology and criminal justice. The
handbook of the criminology of terrorism (pp. 33–46). Wiley-Blackwell: Chichester.
†Weine, S., & Ahmed, O. (2012). Building resilience to violent extremism among Somali –
Americans in Minneapolis: Final report to human factors/behavioral sciences division, science
and technology directorate, U. S. department of Homeland Security.
Wikström, P.-O. H., & Bouhana, N. (2017). Analyzing radicalization and terrorism: A situational
action theory. In G. LaFree & J. D. Freilich (Eds.), The handbook of the criminology of terrorism (pp. 175–186). Chichester: Wiley-Blackwell.
Terrorist Recidivism in Israel: Rates,
Patterns and Risk Factors
Tomer Carmel, Michael Wolfowicz, Badi Hasisi, and David Weisburd
Introduction
The potential for terrorism offenders to re-offend is a key concern for western countries. In the coming years hundreds of terrorism offenders are set to be released in
the US, UK, Europe, and elsewhere. Despite the many advances made in the study
of de-radicalization and desistance, it remains that little is known about terrorist
recidivism. Statistics on terrorist recidivism are hard to come by and are largely
anecdotal. Additionally, little work has been done to examine which factors may
predict terrorist recidivism (Pluchinsky 2008; Silke and Veldhuis 2017).
The study of recidivism generally considers the rate at which released offenders
return to criminality. Depending on the study, recidivism is examined by whether
offenders have been re-arrested, re-convicted, or re-incarcerated. The majority of recidivism research considers offenders to be recidivists regardless of the nature of index
offence or the new offence. That is, involvement in any new offence—regardless of
how it is measured—post-release is considered a marker of recidivism. Such an
approach does not address the specific concerns regarding terrorist recidivism. Rather
than being interested in whether a released terrorism offender may engage in some lowlevel criminal activity, the interest of terrorist recidivism relates to the possibility of
terrorism offenders returning to terrorism. While it has been suggested that we can
learn much about terrorist recidivism and its risk factors from the study of criminal
recidivism, it remains that little work has been done in this area (Altier et al. 2012, 2014).
T. Carmel · M. Wolfowicz (*) · B. Hasisi
Institute of Criminology, Faculty of Law, Hebrew University of Jerusalem, Jerusalem, Israel
e-mail: michael.wolfowicz@mail.huji.ac.il
D. Weisburd
Institute of Criminology, Faculty of Law, Hebrew University of Jerusalem, Jerusalem, Israel
Department of Criminology, Law and Society, George Mason University,
Fairfax, Virginia, USA
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_4
85
86
T. Carmel et al.
The study of terrorist recidivism is therefore the study of recidivism to the same
offence type. However, such study is difficult given the rarity of this type of offending, and as such the low base-rates, as well as the overly long or short observation
periods that are characteristic of terrorism research more broadly. Some studies may
be criticized for examining recidivism over time periods and decades that are disconnected, such as the 1950’s and the post-9/11 era. Others may only include a short
observation period of a few years, which does not allow enough time for offenders
to display recidivism (van der Heide and Schuurman 2018). As a result of these
issues, what little data does exist regarding base-rates for terrorist recidivism is
often questionable.
Like other areas of terrorism research, Israel may be able to serve as an important
case study. Of all democratic countries, Israel suffers the greatest frequency of terrorism. As a result, Israel has incarcerated tens of thousands of ‘security
offenders‘over the last few decades, charged with a variety of violent and non-­
violent terrorism-related offences. In this chapter we analyze data provided by the
Israel Prison Service on all security offenders processed between 2004 and 2017.
The large sample size and long observational period enable us to provide a comprehensive analysis of terrorism-specific recidivism, and key risk factors known to
affect recidivism.
Terrorist Recidivism
The literature from criminology shows that recidivism rates vary greatly depending
on the measure used. Generally speaking, recidivism research looks at re-arrests,
re-convictions, or re-incarcerations for any offence. However, some studies consider recidivism—using one or more of these measure—to the same type of offence
as the most important measure of recidivism (Waldo and Griswold 1979). The study
of recidivism to the same offence type typically relates to the study of offender specialization and versatility.
While it is difficult to summarize this literature, it has generally been found that
offenders are quite versatile (Piquero et al. 2012; Wiesner, Yoerger and Capaldi
2018). However, when comparing recidivism rates for recidivism to the same
offence, there are large differences between different types of offenders. For example, sex offenders, who are often the subject of such inquiry, are often found to
lower levels of specialization and recidivism than other offenders. However, when
disaggregating into rarer offences, such as rape, higher levels of specialization and
recidivism may be found (Miethe, Olson and Mitchell 2006). As Prentky, Lee,
Knight and Cerce (1997) point out in their study of sex offenders over a 25-year
period, had their analysis been limited to the traditional 5-year timeframe, they
would have missed around 20 of the same offence recidivism cases. Similarly, in the
only known study to examine risk factors for terrorist recidivism, Fahey (2013)
found that only time at risk (time since release) was a significant predictor of recidivism. As such there is a need for longer observation periods in the study of recidivism to the same offence for this rare offence type.
Terrorist Recidivism in Israel: Rates, Patterns and Risk Factors
87
Statistics on terrorist recidivism are hard to come by and estimates can vary
widely (Clifford 2018). A recent study of U.S. domestic terrorism offenders found
a recidivism rate of only 1.6% (Hodwitz 2019). Another recent study found that
among terrorism offenders in the Netherlands, between 2008–2012 the recidivism
rate was 4.6% (van der Heide and Schuurman 2018). These numbers are quite close
to the oft cited statistic that only about 2–3% of Irish Republican terrorism offenders released in the Good Friday Agreements recidivated to terrorism related offending (Silke 2014). However, not all estimates are so low, with the most recent
Department of National Security report estimating that 30% of released Guantanamo
Bay detainees had recidivated to terrorism. A similar rate was found among terrorism offenders in Israel between 2008–2012. However, our recent study on Jerusalem
found that the rate was closer to 60% (Hasisi et al. 2019), almost identical to the rate
identified in a recent study of 170 terrorism offenders from the U.S (Altier et al.
2019). Interestingly, these larger rates are quite similar to the estimated rate of
recidivism among German right-wing extremist offenders (Weilnböck, 2012).
Different statistics that are also relevant to the inquiry into terrorist recidivism
can be found among other groups of offenders as well, for example, returning foreign fighters who subsequently were involved in domestic terrorism cases. While it
has been found that among returning foreign fighters in western countries the rate is
about 11% (Hegghammer 2013), the numbers were much higher in Indonesia and
Algeria in the 1980–1990s, at 40 and 90% respectively (Cragin 2017). Similarly,
statistics derived from follow-ups of participants in de-radicalization programs
show that recidivism rates have similar ranges, from 10% in Saudi Arabia to around
50% in both Yemen and Morocco (Hasisi et al. 2019).
There are a number of possible explanations as to why there is a such a wide
range of recidivism rates, seemingly wider than rates for ordinary crime. Recidivism
rates could be high when the specific socio-political environment and context is one
in which extremism is pervasive, where there is a large support base, and where
capable organizations operate (Bélanger 2017; Speckhard, Shajkovci and
Yayla 2017). This may explain why attacks in Northern Ireland subsided after the
“Good Friday Agreement“, whilst they increased in Israel following the Oslo
Accords (Masters 2004). In fact, many of the leaders of the post-Oslo Intifada, and
perpetrators of serious attacks over the next decade were recidivist terrorists (Samuel
1998; Ganor and Falk 2013). High recidivism rates can also be the result of increased
surveillance and policing which give extra attention to released terrorism offenders.
In most countries, several security agencies (including the criminal justice system)
are involved in detecting and thwarting terror activity, much more than effort allocated to cope with criminal activity. Low recidivism rates could therefore be the
result of a lack of surveillance, and the failure of security services to detect recidivism (Bélanger 2017; Porges 2010).
In some cases, low observed recidivism rates may not reflect deterrence but
rather an incapacitation effect, whereby offenders receiving longer sentences have
less of an opportunity to recidivate during the observation period, or are less likely
to recidivate due to their older age at their time of release (Meade et al. 2013; Owens
2009). While this assumption may apply to some terrorism offences, the majority of
terrorism offences are non-violent, and average sentence lengths may be only a few
88
T. Carmel et al.
years.1 Since it is difficult to disentangle the incapacitation and deterrent effects,
prevention efforts should seek to combine incapacitation and deterrent elements to
decrease recidivism to terrorism over time. Saudi Arabia takes such an approach,
where in addition to stiff sentencing, its post-release surveillance has been credited
for its deterrent effect and contribution to low recidivism rates (Hasisi et al. 2019).
The Saudi Arabian case also highlights the role of structures of opportunities for
released terrorists. Whether or not an individual will desist from terrorism or recidivate is dependent on the availability of different opportunities that are more conducive
to desistance, or recidivism. For example, Bjørgo (2008) identifies a range of pull and
push factors that increase the likelihood of desistance. Push factors include societal
and familial disapproval, while pull factors—which are believed to have a stronger
potential influence—include educational, employment and family building opportunities. Individuals may also seek to desist from terrorism when—like crime—they perceive their continued involvement as jeopardizing their extra-­terroristic social and
familial bonds. As such, when policies and programs promote these pull and push
factors post-release, the likelihood of recidivism is reduced. Where these factors are
weakly promoted, or not promoted at all, then the structure of opportunities conducive
to recidivism may be stronger, and thereby the likelihood of recidivism is greater
(Altier et al. 2014). Similar findings have been made with regards to ordinary criminals, and those involved in organized crime in particular (Stys & Ruddell, 2013).
While scholars have theorized that terrorist recidivism rates are likely to be lower
than those for ordinary crime (Silke 2014; Bovenkerk 2011), some have referred to
rates as low as a few percent as ‘suspicious’ (Horgan and Braddock 2010). It is
known that low base rates and inadequate observational periods, which affect much
of terrorism research, impact the development of reliable recidivism rates (Gill et al.
2016). Recidivism rates derived from evaluations of prison-based programs may
also be problematic. These programs often involve volunteer participation with high
motivation to redeem and therefore do not report recidivism on non-participants,
who are often the most radical (Pettinger 2017). As such, these evaluations also do
not provide base-rates of recidivism. Rather, base rates must be developed from
large samples, over longer observation periods, ideally in the absence of any specific rehabilitation programming.
Risk Factors for Recidivism to Terrorism Offending
While Fahey’s (2013) study of Guantanamo Bay detainees, found that only time at
risk (time since release) significantly predicted recidivism, the results highlight the
possibility that terrorist recidivism may be affected by known risk factors for ordiAccording to information presented in the annual Terrorism Situation and Trend Report
(TE-STAT) published by Europol, the average sentence lengths in countries such as the UK,
Germany, and France range from 4–9 years depending on the year of the report. Sentences for terrorism offences are notoriously short in fact (Chesney 2007), and can be made shorter based on the
offence stemming from a sting operation (Norris 2019), or conviction stemming from plead bargaining (De Kerchove and Höhn 2016).
1
Terrorist Recidivism in Israel: Rates, Patterns and Risk Factors
89
nary criminal recidivism. However, while many similarities exist between criminals
and terrorists, there are also important differences and recidivism risk assessment
tools need to be tailored for terrorism offending (Kebbell and Porter 2012; Dernevik
et al. 2009). While the literature on criminal recidivism has identified a range of key
risk factors, it also highlights that such factors may operate differently, or have differential effects depending on the outcome being measured.
In this regard, one of the most commonly examined risk factors is also one of the
most controversial, namely sentence length. There is however a debate in the literature as to whether longer sentence lengths have a deterrent effect or a risk effect, and
the evidence is quite mixed on this issue (Mears, Cochran and Cullen 2015; Rydberg
and Clark 2016; Mitchell et al. 2017; Liem 2013), especially when controlling for
offence type (Rydberg and Clark 2016). It remains unknown as to how sentence
length may affect terrorist recidivism. Sentences for terrorism offences are often
quite short, on account of the fact that most sentences are for non-violent offences,
or include a broad variety of offences (Chesney 2007). However, longer sentence
lengths mean that offenders will be older when they are released, giving them more
time to naturally ‘age out’ of offending (Spivak and Damphousse 2006). This is one
explanation for why longer sentences are often associated with a reduced likelihood
of recidivism. But the effects of aging go beyond this, since as an individual age, life
changes, such as attainment of an academic degree, employment, marriage, and the
baring of children may occur.
Marriage in particular—and children—place new obligations upon an individual, which often include the need to maintain steady employment. However, the
quality of marriage matters. Recent, or weak marriages may not have had the time
to develop into the type of relationship that acts as a protective factor. The quality of
marriage may be especially pertinent in the case of terrorism, since an encouraging
spouse who shares the extremist ideology may increase the likelihood of offending
(Altier, Boyle and Horgan 2019; Silke 2008). Studies of Jihadists have found that
the majority were married, and had children (Bakker 2006; Sageman 2004).
Additionally, Berrebi (2007) found that 45% of Palestinian terrorists were married,
although this number was lower than that of the general population (59%). In a
study of homegrown Jihadists in the US, 20% were found to be married to an
extremist spouse (Klausen et al. 2018). The authors state that “marriage is tied to
status and martyrdom, and it is a frequent precursor to violent action” (p. 4). As
such, marriage may be one factor that operates differently for terrorists and ordinary
criminals (Mullins 2010; Silke 2008).
The high rate of marriage among terrorists may be reflective of the fact that these
offenders tend to be somewhat older than ordinary criminals (Pyrooz, LaFree,
Decker and James 2018). For instance, non-violent terrorism offenders, who make
up the largest group of offenders, tend to be older than violent offenders (Klausen,
Morrill and Libretti 2016; Harms 2017; Perliger et al. 2016). Additionally, Islamic
extremist offenders are more likely to be married than right-wing or left-wing
extremists (Gill, Horgan and Deckert 2014; Chermak and Gruenewald 2015). This
is probably related to the different social & cultural contexts in which they come
from, with more traditional and religious communities encouraging marriage at
younger ages (Silke 2008).
90
T. Carmel et al.
An additional factor that has often been highlighted in the literature is organizational affiliation. For example, the literature on recidivism has found that gang
members are at a heightened risk for recidivism compared to non-gang members
(Dooley et al. 2014). Additionally, many of the recidivists from Guantanamo Bay
were affiliated with known terrorist organizations, where ideology and social network play important role in maintaining the strength and resilience of terrorist
groups. Indeed, Hasisi et al. (2019) found that among terrorism offenders from
Jerusalem, organizational affiliation with one of the major Palestinian organizations
was associated with an increased risk of recidivism. Like in other places, such as the
Philippines, even members of organizations actively involved in peace efforts and
diplomacy may continue to carry out acts of terrorism, and recognized organizations may splinter into new groups, or leave behind semi-autonomous wings. In the
context of the Palestinian organizations, whilst Fatah has formerly renounced terrorism, many factions still flying its banner have refused to do so and continue to
support, promote, and carry out militant activities (Shikaki 2004). Many of its activists have continued to be involved inviolence, including terrorism (Høigilt 2016).
The Current Study
The current study takes advantage of a large dataset of offenders who were incarcerated for terrorism related ‘security offences‘in Israel over a 13-year period
(2004–2017) and examines the characteristics of recidivism to security offending.
There are no de-radicalization or desistance treatment programs for security prisoners in the IPS. Rather, security prisoners were afforded opportunities to engage in
higher education through the Open Universities program.
Methods
Data
The data for this study was obtained from the Israel Prison Service (IPS) who is in
charge of the incarceration of all terrorist offenders, defined as those who “committed an obvious attack on national security or other illegal activities related to
­nationalistic ideology” (Hasisi et al. 2019). These prisoners are classified by the IPS
as “security prisoners” according to the conclusions of the Israeli Security Agency
(ISA). The data included information about all security prisoners who had been
processed for incarceration by the IPS during the years 2004–2017. Prior to 2004,
most security prisoners had been incarcerated by the Israeli Military Police.
However, prison reforms shifted the responsibility for security prisoners to the IPS,
who have since maintained responsibility for all criminal and security prisoners. As
Terrorist Recidivism in Israel: Rates, Patterns and Risk Factors
91
such, from 2004 onwards all data collection, storage and processing for both security and criminal offenders was placed under a centralized and uniform system. To
ensure consistency of the data, this study included only those prisoners who were
incarcerated during or after 2004 and offenders whose first incarcerations occurred
prior to 2004 were removed from analysis.
The data included both incarceration and individual level risk factors that are
known to affect criminal recidivism: age at first incarceration, age at time of release,
length of full sentence as well as length of actual incarceration, terrorist organization affiliation and marital status. In addition, offence type classification for each
incarceration was coded and included: public disorder, assault, murder and manslaughter and national security. National security offences generally relate to
offences such as possession/manufacture of weapons, possession of weapons, affiliation with proscribed organizations, or contact with foreign agents. Public disorder
offences most commonly related to violent rioting and other forms of low intensity
violence such as rock throwing, road blocking etc. Other types of crimes, such as
property, sex and fraud, were categorized as “other”.
The dataset consisted of 26,667 individual offenders that had been incarcerated
and convicted for at least one security offence. Of these, 25,253 were from the West
Bank, 1414 were from Jerusalem. Combined they were imprisoned for a total of
31,335 terms, with an average of 1.18 incarcerations per individual. 27 of the prisoners in the sample were female, and 37 were Jewish. Since none of the Jewish or
female prisoners that were released prior to data collection were re-incarcerated for
security offenses, they were not included in the analysis.
Variables
The study aimed to test the impact risk factors that are known to affect criminal
recidivism by examining the re-incarceration of security prisoners for terrorism
related charges after release. Most of the sample lives under the jurisdiction of the
Palestinian Authority. When approached by the researchers, the PA law enforcement agency refused to give any information regarding the criminal activity of subjects included in our sample. Because the information regarding criminal activity of
the majority of the sample was incomplete, it was excluded also from analysis.
History of prior security convictions Having a history of security incarcerations
greatly affects the risk of recidivism (Hasisi et al. 2019). Hence, the number of prior
incarcerations as a security prisoner was included in this study as well. It was impossible for the researchers to obtain information regarding the criminal history of
security prisoners that live under the jurisdiction of the Palestinian Authority. Since
most of our sample might have missing criminal records, any ordinary criminal
incarcerations in the sample were removed from analysis, and the effects criminal
history were not explored here.
92
T. Carmel et al.
Age at time of release The IPS data includes the release date of each incarceration
as well as the prisoner’s date of birth. As such, we calculated age at time of release
based on this data. To disentangle age and history of incarceration, age at time of
release was tested only for first incarceration.
Length of sentence served Similar to the length of the sentence, length of sentence
served is a calculation of the number of weeks the prisoner was held in prison.
Unlike the length of sentence, that is strictly based on the ruling of the court, the
length of sentence served is based on the actual term of imprisonment. The length
of the sentence served cannot be longer than the length of the sentence ruled. As
such, both correlate. Yet, from time to time, agreements between the Israeli government and the Palestinian Authority or the Hamas can result in release of security
prisoners before their term. A major event, when a thousand prisoners were released
at once, occurred during the period of the study. Thus, correlation between time
sentenced and time served is not as strong as one might expect.
Offence type Offence type was categorized according to the index used by the
IPS and Israel Police. Categories used were national security, public disorder,
assault, murder and manslaughter, sex, vice and property. National security
offences generally relate to offences such as possession/manufacture of weapons,
possession of weapons, affiliation with proscribed organizations, or contact with
foreign agents. Public disorder offences most commonly related to violent rioting
and other forms of low intensity violence such as rock throwing, road blocking etc.
Crimes that did not fit into one of the above categories were coded as “other”. One
incarceration might relate to several types of crime, whether since the act included
more than one offense or whether the incarceration is due to several independent
acts. Hence, each category was indexed using a dummy variable, to account for
this scenario.
Organizational affiliation Many security prisoners are known to be affiliated with
one of the primary organizations linked to the Israeli-Palestinian conflict. Such affiliation was provided by the IPS, based on the identification done by the prisoners
themselves and assessment of the Israeli Security Agency.
Marital status During the time of the study, the IPS was provided with information about the marital status of the prisoners by the prisoners themselves. Although
the information is self-reported, it is regarded as reliable, since it reflects on the
rights of the immediate family of the prisoner for visitations. In many cases in the
sample, the family lives under the jurisdiction of the Palestinian Authority and is
required to be registered and authorized for visitation, contributing further for the
validity of the information regarding the prisoners’ immediate family. In order to
protect the privacy of the prisoners, the IPS provided marital status in three categories only: Single/never married, married, or divorced or widowed. Prisoners
who refused to provide marital status (7.2% of the sample) were categorized as
“unknown”.
93
Terrorist Recidivism in Israel: Rates, Patterns and Risk Factors
Analysis
Overall recidivism rates were assessed using the Kaplan-Meier method (Kaplan and
Meier 1958). For the effects of risk factors over terrorist recidivism a specific cutoff
of recidivism after 5 years from release was used. As recidivism rates in Israel are
slow to reach an asymptote, the IPS recommends using 5-years recidivism as a
yardstick for measuring recidivism (Ben-Zvi and Walk 2011).
Results
Recidivism Rates
The sample demonstrated a relatively modest recidivism rate, reaching 6.8%, 14.1%
and 17.3% 1-year, 3-year and 5-year rates, respectively (Fig. 1). The shape of the
recidivism curve is a logarithmic function that follows the general pattern for ordinary criminal recidivism, although recidivism of security prisoners was lower than
general criminal recidivism in Israel (Walk and Berman 2015).
History of Prior Security Convictions
Similar to the criminal recidivism literature, prior incarcerations for terrorism-­
related offences increase the risk of recidivism. While only 16.3% of prisoners without prior incarcerations recidivate 5 years after release, the rate is almost double for
Recidivism to security incarceration
25%
20%
15%
10%
5%
0%
0
1
2
3
4
5
6
Years after release
Fig. 1 Overall recidivism rates by year after release
7
8
9
10
94
T. Carmel et al.
Recidivism to security offences
40%
35%
30%
Prior security
incarcerations
25%
2
20%
1
15%
0
10%
5%
0%
0
2
4
6
8
10
Years after release
Fig. 2 Recidivism rates by year after release according to number of prior incarcerations as security prisoners
those with 2 prior incarcerations, at 30.4%. Similarly, prisoners with prior incarcerations tend to be re-incarcerated faster than those without such history. These
patterns are similar to what is observed in ordinary criminal recidivism (Fig. 2).
Age at Time of Release
Age at time of release ranged between 14.5 and 68.8 years of age, averaging at 24
(SD = 6.33). 11.9% of the sample were legally minor, being younger than 18 years
old, whilst only 6.2% of the sample were older than 35 years old. As depicted below
(Fig. 3), recidivism to another security incarceration was highest for those released
at a younger age. The rate dropped significantly with age, reaching an asymptote as
early as age 27.
Age at First Incarceration
Since the majority of the sample were not recidivists, and only had one incarceration, age at first incarceration was very similar to age at release, averaging at 21.9
(SD = 5.53). The trend also follows a similar pattern to age at release. Most of the
sample were first incarcerated as juveniles (17.9% of the sample) or young adults
(79.2% of the sample), and only 2.9% were incarcerated over the age of 35 (see
Fig. 4).
95
Terrorist Recidivism in Israel: Rates, Patterns and Risk Factors
Recidivism to another security
incarceration
45%
40%
35%
30%
25%
20%
15%
10%
5%
0%
15
20
25
30
35
40
45
50
55
60
Age at time of release
Fig. 3 5-year recidivism to security incarceration by age at time of release. Error bars represent
standard errors
Recidivism to security incarceration
50%
45%
40%
35%
30%
25%
20%
15%
10%
5%
0%
15
20
25
30
35
40
45
50
Age at first incarceration
Fig. 4 5-year recidivism to security incarceration by age at first incarceration. Error bars represent
standard errors
Length of Sentence Served
In our sample, the length of sentence served had a complex impact on recidivism
(Fig. 5). The length of sentences of up to 7 years reduced recidivism, while the
effect of sentences longer than 7 years did not continue this trend. Although
­sentences longer than 7 years show a slight increase in recidivism compared to the
7 years long sentences, recidivism maintains under the high rates of short sentences.
T. Carmel et al.
25%
30%
Recidivism to security incarceration
Recidivism to security incarceration
96
25%
20%
15%
10%
5%
0%
0
13
26
39
52
65
78
91 104
Length of served sentence (weeks)
Up to two years
20%
15%
10%
5%
0%
2
4
6
8
10
Length of served sentence (years)
Fig. 5 5-year recidivism to security incarceration by length of sentence served up to 2 years (left
panel, in weeks) and over 2 years (right panel, in years)
Table 1 Recidivism by type
of current offense. Specific
incarcerations can be counted
several times due to being
related to more than one
offense type
Offence type
N
National security
24,536
Murder/manslaughter
9222
Public disorder
1906
Assault
578
Property
546
Recidivism
rate
17.1%
18.4%
25.1%
25.8%
27.6%
The decline in recidivism by length of sentence served is evident also in the relatively short sentences of up to 2 years (Fig. 5, left panel).
Offence Type
As noted above, offence type, and the type of offences contained in offender histories, are known predictors of recidivism. Accordingly, offence types affected the
risk of recidivism in our sample (Table 1). Following findings from criminology, the
recidivism rate for offenders who committed violent (assault offense) and disorderly offences--(25.7%--was found to be significantly higher than the rate found for
offences tagged as “national security”, which include offences such as p­ ossession/
manufacture of weapons, possession of weapons, affiliation with proscribed organizations, or contact with foreign agents, were correlated with significantly lower
recidivism (Χ2(1) = 134.29, p < 0.0001).
Terrorist Recidivism in Israel: Rates, Patterns and Risk Factors
Table 2 Number of
individuals affiliated to each
Palestinian organization,
and their respective
recidivism rate
Table 3 Prevalence by
marital status and respective
recidivism rates
97
Affiliation
N
Unaffiliated
3733
Fatah
14,105
Hamas/Islamic Jihad
8813
Secular factions
1701
Status
Single
Married
Divorced or widowed
Unknown
Prevalence
78.0%
17.9%
<0.1%
4.1%
Recidivism
rate
8.2%
18.3%
18.6%
22.1%
Recidivism
18.8%
15.4%
12.3%
3.2%
Affiliation
Most individuals in our samples were affiliated with one of the major Palestinian
organizations. About half (49.7%) was affiliated with Fatah and a third (31.1%) with
Hamas and Islamic Jihad (31.1%). 6.0% were affiliated with one of the secular
organizations (e.g. Popular Front for the Liberation of Palestine), and a negligible
number (<0.1%) were affiliated with “other” groups. 13.2% of the prisoners were
unaffiliated. Whilst recidivism rates for affiliates of the different organizations were
nearly identical (see Table 2), they were significantly greater than unaffiliated
(Χ2(1) = 247.7, p < 0.0001).
Marital Status
The majority of the sample were registers as being single, whereas 17.9% were married, <0.1% divorced or widowed, and 4.1% were listed with an unknown marital
status. Overall, the recidivism rate for offenders listed as single was significantly
greater that for all other marital statuses (Χ2(1) = 31.87, p < 0.0001), while the prisoners that did not report marital status only seldom recidivated (Table 3).
Discussion
A growing literature that approaches terrorism from a criminological perspective
has now identified that significant overlaps exist between criminals and terrorism
offenders, as well as patterns in criminal and terrorism related phenomena. In this
chapter we set out to identify to what extent such overlaps may extend to recidivism,
namely with respect to the rates of recidivism and recidivism-related risk factors.
98
T. Carmel et al.
Our study in the Israeli context represents an important case study that could have
implications for other settings. A study of recidivism in Israel based on data from
2008 found that the 5 year recidivism rate for security offenders was 7.9%, and
41.3% for ordinary crime (Walk and Berman 2015). On the other hand, our prior
research on a sub-sample of terrorism offenders from Jerusalem we found that the
five-year terrorist recidivism rate was 60.2% (Hasisi et al. 2019). These differences
can be explained by the fact that the population of security prisoners in Israel is
predominantly not from Jerusalem but from the West Bank (about 95%), hence our
results are not contradictory. Our findings of a recidivism rate of 20%, being near
exactly half of the recidivism rate for ordinary crime in Israel, demonstrate an
important overlap with the relative rates for recidivism to the same offence noted
above. They also demonstrate that recidivism rates deduced from limited observational periods may underestimate the phenomenon (Walk and Berman 2015).
In this regard, our study highlights the importance of looking at recidivism over
a longer period of time. Had we limited our analysis to 5 years, for those with 2 or
more prior incarcerations, we would have missed out on 5% of the recidivists in the
dataset. Following general trends observed in criminal recidivism, we found that the
rate of recidivism increased following each additional incarceration for a terrorism
offence. Since each additional incarceration involves a period in which the offenders is incapacitated and unable to re-offend, the importance of the longer observation period is even more striking. In fact, the small differences in recidivism between
5 and 10 years, relative to sentences less than 5 years, indicate the existence of an
incapacitation effect in determining recidivism rates.
These findings are inherently dependent on the effects of incarceration length. As
discussed in the literature review, the evidence is quite mixed as to whether longer
incarcerations have a greater deterrent or criminogenic effect when it comes to recidivism. In the context of terrorism, much of the focus has been on the role of longer
incarcerations. However, as demonstrated by our data, the majority of terrorism-­
related offences may be for non-violent offences, or at least offences that did not lead
to the actual commission of a successful terror attack. As such, like in Europe and the
US, incarcerations for terrorism-related offences can actually be quite short.
In this study, a clear decline in recidivism was associated with longer sentences
up to 7 years long. However, the trends seem to be less conclusive in sentences longer than 7 years. As has been noted in the literature, it can be quite difficult to disentangle the degree to which incapacitation and deterrence contribute to declining
recidivism rates associated with longer incarceration periods. One of the issues with
shorter sentences is that offenders are still young when they are released, and
thereby still within the peak ‘risk’ period for offending. On the other hand, longer
sentences are often associated with older age at the time of release, meaning that the
incapacitation period has given the offenders an opportunity to ‘age out’ of offending. Given the mean age of our sample fell within the peak risk period, relatively
short sentences of a few years might be enough to incapacitate high-risk prisoners
in a critical period in their life (see Figs. 2 and 3 for support of this assumption).
Possible deterrent effects can potentially be deduced by the declining rates of
recidivism associated with offenders serving even exceptionally short incarceration
Terrorist Recidivism in Israel: Rates, Patterns and Risk Factors
99
periods. In this range of incarceration lengths, the age of offenders does not change
dramatically, thus it is unlikely that mere incapacitation is the cause of the declining
trend. While it is clear that both incapacitation and deterrence play a role, our data
does not permit for extrapolating to what degree these two causes may be responsible for the observed trends.
With regards to offence type, can be said to be consistent with the evidence from
research on criminal recidivism. Generally speaking, violent offenders have higher
rates of recidivism than non-violent offenders. However, this may exclude both violent and non-violent offenders of rare crimes, for which recidivism rates are also
lower (Blokland and Nieuwbeerta 2004). Our results show that more severe
offences, such as manslaughter and murder, were associated with relatively low
recidivism rates. This is understandable given that such offences are generally associated with longer incarceration periods. As discussed, longer sentences reduce the
amount of time during the observation period in which the offender has an opportunity to recidivate, and also means that offenders are older at the time of their release,
which in and of itself is associated with lower recidivism as well.
Perhaps most importantly however is our findings that offenses classified as
“national security” were associated with the lowest recidivism rates. This offense
category includes a wide range of offenses, from being an active member of an
illegal organization to producing weapons, and coordinating attacks. It can be said
that offences in this category are generally less descriptive of direct confrontational
and violent behavior. It is possible that prisoners who were more active in participating in violent acts will be more prone to being involved again after their release
than those who were less active in the first place. Nevertheless, given how general
this offence category is, it is difficult to try to extrapolate the underlying explanations for the low recidivism rates.
As our data highlights, the majority of offenders were affiliated with a recognized terrorist organization. Unaffiliated prisoners were significantly less likely to
recidivate than affiliated prisoners. Among affiliated offenders, there were little differences in the recidivism rates, regardless of which organization they were affiliated with; although these organizations differ in considerably in terms of their
identity and doctrine (e.g. religious, secular). It can generally be said that these
organizations enjoy relatively high popular support. It is possible that the similarities in recidivism rates associated with affiliated offenders reflects such support.
The role of such support may also be considered with respect to marital status,
for which our findings are similar to what has generally been found in the literature
on ordinary criminal recidivism. Namely, that married individuals are less likely to
recidivate than non-married. However, these results stand in contrast to previous
findings that marriage can be a risk factor for terrorist recidivism (Altier et al. 2019;
Hasisi et al. 2019). Additionally, it is interesting to note that among the small sample
of offenders for whom a marital status identification is lacking, the rate of recidivism was found to be exceptionally low (3.2%). Since the reason for the missing
information is difficult to assess, it is difficult to explain why they had such a low
rate of recidivism. Further investigation of the characteristics of prisoners with
missing marital status is needed to resolve this puzzling outcome. Given the dis-
100
T. Carmel et al.
crepancies, further research is needed in order to disentangle the risk-protective
effects of marriage on recidivism to terrorism. Whilst every country’s experience
with terrorism is unique, Israel has often served as an important case study of terrorism and counter-terrorism that may be applicable to other Western contexts.
Lastly we identified important differences between recidivism rates of offenders
from Jerusalem and the West Bank. Similar to the effects of organizational affiliation and marriage, these findings can possibly be explained by differences in the
structures of support, but also of opportunities that exist between these two places
(Bélanger 2017; Speckhard et al. 2017). Firstly, Jerusalem residents reside in areas
under full Israeli administration. Their incarceration histories mean that their
employment opportunities may be more restricted. Additionally, they may be subject to more ongoing interaction with the Israeli state and with security services in
particular. On the other hand, West Bank residents may be able to live their lives in
the Palestinian Authority governed territories quite separate from the Israeli state
and its security apparatus. Here, the general society would view them as having
‘served their time’ for a national goal. Released terrorism offenders be offered better employment opportunities, with many jobs, especially government jobs under
the auspices of the Palestinian Authority, giving preference to, and in some cases
being reserved for those who have spent time in Israeli prisons for security-related
offences.2 Unlike ordinary criminals, terrorism offenders are viewed as legitimate,
and rather than being labeled as deviant, their activities may lead them to gain
higher social status. As such, the opportunities in the West Bank, which has a­ rguably
witnessed much improvement compared to East Jerusalem, may be more conducive
to desistance.3
As a large literature emphasizes, whilst every country’s experience with terrorism is unique—and the Israel and Palestinian one is no different in this regard—
important lessons can be drawn from seemingly different contexts. In this regard,
Israel has often served as an important case study for lessons on terrorism and
counter-­terrorism that may be applicable to other Western contexts. As described in
the literature review, there is significant overlap in the existing evidence pertaining
to trends in terrorism-related phenomena. A growing literature that approaches terrorism from a criminological perspective has now identified that significant overlaps exist between criminals and terrorism offenders, as well as patterns in criminal
and terrorism related phenomena. In this study we have identified that these overlaps also extend to the observed patterns of recidivism and the different factors that
affect these patterns.
While the Palestinian Authority sometimes denies such claims, the Arabic website for the
Commission of Detainees and ex-Detainees Affairs includes links to laws and regulations concerning payments to detainees and former detainees, as well as exclusive job advertisements which
include being a former detainee as a pre-requisite for application. See http://cda.gov.ps/
3
Polls conducted by the Palestinian Center for Policy and Survey Research (PCPSR) consistently
show that support for violence and armed struggle is consistently greater in “villages/towns/cities”
as well as “refugee camps” than it is in “the West Bank”. The results of the polls can be found at
http://pcpsr.org/en/node/154.
2
Terrorist Recidivism in Israel: Rates, Patterns and Risk Factors
101
Conclusions
The study of terrorist recidivism has to date not received the type of attention that it
deserves and demands. In the coming years, hundreds of terrorism offenders are set
to be released in Western countries, especially in Europe and the U.S. While many
commentators and analysts have expressed fears over the potential for these offenders to re-offend, little is known about what the actual likelihood of recidivism is. To
date, most studies have focused primarily on qualitative assessments of programs
that seek to promote de-radicalization and disengagement, by which one measure is
recidivism. However, with a lack of data on what base-line rates of terrorist recidivism may look like, and what factors may affect terrorist recidivism, it is difficult
for programs to be evaluated based on recidivism measures. Additionally, not all
terrorism offenders are participating in such programs.
The study presented in this chapter suggests that like many other terrorism-­
related phenomenon, terrorist recidivism is likely to follow patterns that closely
resemble those found with respect to ordinary crime. Like other forms of recidivism
to the same offence type for rare offences, terrorist recidivism can be expected to
reach approximately half of the general recidivism rate. Similarly, risk factors such
as age at time of incarceration and release, number of prior incarcerations, and
length of incarceration, are likely to impact the likelihood of recidivism to terrorism
in a similar fashion as they do for ordinary criminal recidivism.
While the study presented in this chapter provides additional evidence to support
the continued application of criminological perspectives to the study of terrorism,
and highlights additional areas of overlap between criminal and terrorism offenders,
it also highlights that key differences also exist. Given these findings, it is imperative that researchers further investigate the role of factors known to affect recidivism
with respect to terrorist recidivism, in order to better develop risk assessment
approaches, and evaluations of programs that aim to prevent the re-offending of
extremist offenders following their release from prison.
References
Altier M. B., Horgan, J., Thoroughgood, C. (2012). Returning to the fight: what the literature on
criminal recidivism can contribute to our understanding of terrorist recidivism. Department of
Homeland Security report
Altier, M. B., Thoroughgood, C. N., & Horgan, J. G. (2014). Turning away from terrorism: Lessons
from psychology, sociology, and criminology. Journal of Peace Research, 51(5), 647–661.
Altier, M. B., Leonard Boyle, E., & Horgan, J. G. (2019). Returning to the fight: an empirical
analysis of terrorist reengagement and recidivism. Terrorism and Political Violence, 1–25.
Bovenkerk, F. (2011). On leaving criminal organizations. Crime, Law and Social Change, 55(4),
261–276.
Blokland, A. A. J., & Nieuwbeerta, P. (2004). Recidive en het beëindigen van de criminele carrière
over een periode van 25 jaar. Tijdschrift voor Criminologie, 46(1), 18–36.
Ben-Zvi, K., & Walk, D. (2011). Recidivism shel asirim plilim m’shuchrarey 2004 b’Yisrael
[Recidivism of criminal prisoners released in 2004 in Israel]. Tzohar l’Beit ha’Sohar [Window
into the Prison], 14, 10–28.
102
T. Carmel et al.
Berrebi, C. (2007) Evidence about the Link Between Education, Poverty and Terrorism among
Palestinians. Peace Economics, Peace Science and Public Policy 13(1).
Bakker, E. (2006). Jihadi terrorists in Europe. The Hague: Cliengendael.
Bjørgo, T. (2008). Processes of disengagement from violent groups of the extreme right. In Leaving
terrorism behind (pp. 48–66). Routledge.
Bélanger, J. J. (2017). The rise and fall of violent extremism: The science behind community-­
based interventions. In The Motivation-Cognition Interface (pp. 170–195). Routledge.
Clifford, B. (2018). Radicalization in Custody. Towards Data-Driven Terrorism Prevention in the
United States Federal Correctional System, Program on Extremism Policy Paper, The George
Washington University.
Cragin, R. K. (2017). The challenge of foreign fighter returnees. Journal of contemporary criminal
justice, 33(3), 292–312.
Chermak, S., & Gruenewald, J. A. (2015). Laying a foundation for the criminological examination
of right-wing, left-wing, and Al Qaeda-inspired extremism in the United States. Terrorism and
Political Violence, 27(1), 133–159.
Chesney, R. M. (2007). Federal prosecution of terrorism-related offenses: conviction and sentencing data in light of the soft-sentence and data-reliability critiques. Lewis & Clark L. Rev.,
11, 851.
Dernevik, M., Beck, A., Grann, M., Hogue, T., & McGuire, J. (2009). The use of psychiatric
and psychological evidence in the assessment of terrorist offenders. The Journal of Forensic
Psychiatry & Psychology, 20(4), 508–515.
Dooley, B. D., Seals, A., & Skarbek, D. (2014). The effect of prison gang membership on recidivism. Journal of Criminal Justice, 42(3), 267–275.
De Kerchove, G., & Höhn, C. (2016). “The Regional Answers and Governance Structure
for Dealing with the Foreign Fighters: The Case of the EU”, In De Guttry, A.; Capone, F.;
Paulussen, C. (Eds.), Foreign Fighters under International Law and Beyond, T.M.C. Asser
Press/Springer Verlag: The Hague, 299–332.
Fahey, S. (2013). Predictors of Release from Guantánamo Bay and Detainee Recidivism.
International Journal of Criminology and Sociology, 2, 453–468.
Ganor, B., & Falk, O. (2013). De-radicalization in Israel’s prison system. Studies in conflict &
terrorism, 36(2), 116–131.
Gill, P., Horgan, J., & Deckert, P. (2014). Bombing alone: Tracing the motivations and antecedent
behaviors of lone‐actor terrorists. Journal of Forensic Sciences, 59(2), 425–435.
Gill, P., Horgan, J., Corner, E., & Silver, J. (2016). Indicators of lone actor violent events: The
problems of low base rates and long observational periods. Journal of Threat Assessment and
Management, 3(3–4), 165.
Harms, J. (2017). The war on terror and accomplices: An exploratory study of individuals who
provide material support to terrorists. Security Journal, 30(2), 417–436.
Hodwitz, O. (2019). The Terrorism Recidivism Study (TRS). Perspectives on Terrorism,
13(2), 54–64.
Hasisi, B., Carmel, T., Weisburd, D., & Wolfowicz, M. (2019). Crime and terror: examining criminal risk factors for terrorist recidivism. Journal of Quantitative Criminology, 1–24.
Horgan, J., & Braddock, K. (2010). Rehabilitating the terrorists?: Challenges in assessing the
effectiveness of de-radicalization programs. Terrorism and Political Violence, 22(2), 267–291.
Høigilt, J. (2016). Fatah from below: The clash of generations in palestine. British Journal of
Middle Eastern Studies 43(4), 456–471.
Hegghammer, T. (2013). Should I Stay or Should I Go? Explaining Variation in Western
Jihadists' Choice between Domestic and Foreign Fighting. American Political Science Review
107(1), 1–15
Klausen, J., Morrill, T., & Libretti, R. (2016). The terrorist age‐crime curve: An analysis of
American Islamist terrorist offenders and age‐specific propensity for participation in violent
and nonviolent incidents. Social Science Quarterly, 97(1), 19–32.
Terrorist Recidivism in Israel: Rates, Patterns and Risk Factors
103
Kebbell, M. R., & Porter, L. (2012). An intelligence assessment framework for identifying individuals at risk of committing acts of violent extremism against the West. Security Journal,
25(3), 212–228.
Kaplan, E. L., & Meier, P. (1958). Nonparametric estimation from incomplete observations.
Journal of the American statistical association, 53(282), 457–481.
Klausen, J., Libretti, R., Hung, B. W., & Jayasumana, A. P. (2018). Radicalization trajectories:
An evidence-based Computational approach to dynamic risk assessment of “Homegrown”
Jihadists. Studies in Conflict & Terrorism, 1–28.
Liem, M. (2013). Homicide offender recidivism: A review of the literature. Aggression and Violent
Behavior 18(1), 19–25.
Miethe, T. D., Olson, J., & Mitchell, O. (2006). Specialization and persistence in the arrest histories of sex offenders: A comparative analysis of alternative measures and offense types. Journal
of Research in Crime and Delinquency, 43(3), 204–229.
Mears, D. P., Cochran, J. C., & Cullen, F. T. (2015). Incarceration heterogeneity and its implications for assessing the effectiveness of imprisonment on recidivism. Criminal Justice Policy
Review, 26(7), 691–712.
Mitchell, O., Cochran, J. C., Mears, D. P., & Bales, W. D. (2017). Examining prison effects on
recidivism: A regression discontinuity approach. Justice Quarterly, 34(4), 571–596.
Mullins, S. (2010). Rehabilitation of Islamist terrorists: Lessons from criminology. Dynamics of
Asymmetric Conflict 3(3), 162–193.
Masters, D. (2004). Support and Nonsupport for Nationalist Rebellion: A Prospect Theory
Approach. Political Psychology 25(5), 703–726.
Meade, B., Steiner, B., Makarios, M., & Travis, L. (2013). Estimating a dose–response relationship
between time served in prison and recidivism. Journal of Research in Crime and Delinquency,
50(4), 525–550.
Norris, J. J. (2019). Explaining the Emergence of Entrapment in Post-9/11 Terrorism Investigations.
Critical Criminology 27(3), 467–483.
Owens, E. G. (2009). More Time, Less Crime? Estimating the Incapacitative Effect of Sentence
Enhancements. The Journal of Law and Economics 52(3), 551–579.
Pluchinsky, D. A. (2008). Global jihadist recidivism: A red flag. Studies in Conflict & Terrorism,
31(3), 182–200.
Prentky, R. A., Lee, A. F., Knight, R. A., & Cerce, D. (1997). Recidivism rates among child
molesters and rapists: A methodological analysis. Law and Human Behavior, 21(6), 635–659.
Pyrooz, D. C., LaFree, G., Decker, S. H., & James, P. A. (2018). Cut from the same cloth? A
comparative study of domestic extremists and Gang Members in the United States. Justice
Quarterly, 35(1), 1–32.
Pettinger, T. (2017). De-radicalization and Counter-radicalization: Valuable Tools Combating
Violent Extremism, or Harmful Methods of Subjugation?. Journal for Deradicalization,
(12), 1–59.
Perliger, A., Koehler-Derrick, G., & Pedahzur, A. (2016). The gap between participation and violence: Why we need to disaggregate terrorist ‘profiles’. International Studies Quarterly, 60(2),
220–229.
Porges, M. L.(2010). Deradicalisation, the Yemeni Way. Survival 52(2), 27–33.
Piquero, A. R., Jennings, W. G., & Barnes, J. C. (2012). Violence in criminal careers: A review of
the literature from a developmental life-course perspective. Aggression and Violent Behavior,
17(3), 171–179.
Rydberg, J., & Clark, K. (2016). Variation in the incarceration length-recidivism dose–response
relationship. Journal of Criminal Justice, 46, 118–128.
Silke, A., & Veldhuis, T. (2017). Countering violent extremism in prisons: A review of key recent
research and critical research gaps. Perspectives on Terrorism, 11(5), 2–11.
Speckhard, A., Shajkovci, A., & Yayla, A. (2017). Following a military defeat of ISIS in Syria
and Iraq: what happens next after the military victory and the return of foreign fighters?.
Contemporary Voices: St Andrews. Journal of International Relations, 8(1), 81–89.
104
T. Carmel et al.
Sageman, M. (2004). Understanding terror networks. University of Pennsylvania Press,
Philadelphia.
Spivak, A. L., & Damphousse, K. R. (2006). Who returns to prison? A survival analysis of recidivism among adult offenders released in Oklahoma, 1985–2004. Justice Research and Policy,
8(2), 57–88.
Silke, A. (2008). Research on terrorism. In Chen H., Reid E., Sinai J., Silke A., Ganor B. (eds)
Terrorism Informatics. Integrated Series In Information Systems, vol 18. Springer, Boston,
MA (pp. 27–50).
Silke, A. (2014). Risk assessment of terrorist and extremist prisoners. In A. Silke (Eds.) Prisons,
terrorism and extremism: Critical issues in management, radicalisation and reform, 108–121.
Routledge. New York.
Stys, Y., & Ruddell, R. (2013). Organized crime offenders in Canada: Risk, reform, and recidivism. Journal of Offender Rehabilitation, 52(2), 75–97.
Samuel, E. B. (1998). Israel's Demoralization. Middle East Quarterly 5(3), 3–12.
van der Heide, L., & Schuurman, B. (2018). Reintegrating Terrorists in the Netherlands: Evaluating
the Dutch approach. Journal for Deradicalization, 17, 196–239.
Waldo, G., & Griswold, D. (1979). Issues in the measurement of recidivism. Sechrest, L., White,
S.O. and Brown, E.D. (Eds.), The rehabilitation of criminal offenders: Problems and prospects,
225–250. Washington, DC: The National Academies Press
Wiesner, M., Yoerger, K., & Capaldi, D. M. (2018). Patterns and correlates of offender versatility and specialization across a 23-year span for at-risk young men. Victims & Offenders,
13(1), 28–47.
Walk, D., Berman, E. (2015). The Recidivism of Israeli Prisoners. Israel Prison Service, Ramla
Weilnböck, H. (2012). Hate Crime and Radicalisation: the German Political Experience and the
Schellenberg Report. Belfast: NIACRO.
Unintended Negative Outcomes
of Counter-Terrorism Policing: Procedural
(in)Justice and Perceived Risk
of Recruitment into Terrorism
Justice Tankebe
Introduction
Although terrorism has always attracted academic interests, these interests have
grown exponentially since the New York terrorist attacks of September 2001.
Several cities have since suffered similar attacks, including London (UK), Bagdad
(Iraq), Madrid (Spain), Bali (Indonesia), Mumbai (India), and Konduga (Nigeria)
(see http://apps.start.umd.edu/gtd/). The rise and demise of the Islamic State of Iraq
and Syria (ISIS) further increased the global research and policy interest in the
topic. Among the key research questions is the risk of radicalisation, or recruitment
into terrorism (Borum 2011; Dalgaard-Nielsen 2010). Radicalisation refers to “a
process of escalation from nonviolent to increasingly violent repertoires of action
that develops through a complex set of interactions unfolding over time” (della
Porta 2018: 461). The research evidence points to the role of various structural and
individual factors, including criminal history, lack of employment opportunities and
mental health (LaFree et al. 2018; McCauley and Moskalenko 2008).
However, a major cautionary tale in counter-terrorism practices is the risk of
unintended negative effects; that is, of these practices increasing rather than reducing the risk of terrorism. Thus, 2 years after the terrorist attacks in New York on
September 2001, Mr Kofi Annan, the late Secretary General of the United Nations,
remarked at a conference that counter-terrorism practices risked “acting as a recruiting sergeant for the very terrorists we seek to suppress” (Annan 2003). Unintended
negative outcomes of counter-terrorism may also find expression in public disinclination to cooperate with security agencies, such as refusing voluntarily to supply
them with information about terrorist activities in their local communities. The concern with unintended negative outcomes arises from the perspective that terrorism
is not merely a dyadic interaction between states and violent non-state actors; on the
J. Tankebe (*)
Institute of Criminology, University of Cambridge, Cambridge, UK
e-mail: jt340@cam.ac.uk
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_5
105
106
J. Tankebe
contrary, terrorist campaigns involve a triad of “strategic actors – the group, the
government, and the audience” Cronin (2009). A feature of the dynamic relationship between the State and terrorist groups, argues Cronin, is a struggle to claim
legitimacy among local communities. On this perspective, terrorists sometimes
tempt the State into actions that delegitimate it both in the eyes of domestic and
international audiences (Cronin 2009).
Understanding how people experience and perceive counter policing and its
impact can, therefore, be vital not only in encouraging public collaboration, but in
preventing radicalisation (see Murphy et al. 2017; Mumpower et al. 2013). Counter-­
terrorism practices take different forms and at different levels, and involve a multiplicity of actors. One of these actors are the police, and Mr. Annan’s caution draws
attention to the nature of the police’s counter-terrorism practices. In the United
Kingdom, counter-terrorism strategies are believed to have led to the singling out of
sections of populations for “state attention as being ‘problematic’ […and] may be
targeted, not necessarily as a result of suspected wrong doing, but simply because of
their presumed membership to that sub-group” (Pantazis and Pemberton 2009:
647). Counter-terrorism strategy in the United Kingdom – especially the element of
prevent, which aims to “stop people becoming terrorists or supporting terrorism”
(Home Office 2018: 8) – has been more controversial because of concerns that it
disproportionately targets Muslim communities. Consequently, scholars have raised
concerns that the strategy has made Muslims a “suspect community”, with the
attendant fear that the nature of their everyday interactions with police raises questions about unintended, even counter-productive, consequences (e.g. Choudhury
and Fenwick 2011).
These concerns resonate with the arguments of police legitimacy theory
according to which “the public evaluates police, courts, and the law primarily in
terms of how authority is exercised. Police build perceived legitimacy among
the public by treating people fairly during personal encounters” (Tyler et al.
2010: 370). Research evidence from nonterror shows that perceived fairness of
police practices changes people’s orientations towards the law, including
increased likelihood of offending. However, with a few exceptions (for example, Murphy, Cherney and Teston 2019; Tyler et al. 2010), this literature has not
been extended to terrorism. Moreover, studies such as Tyler et al. (2010) rely on
correlational designs. Thus, the results reported in this chapter extends current
literature in two important ways: first, we apply legitimacy theory to test the
claims about unintended outcomes of counter-terrorism practices. Specifically,
and drawing upon data from an experimental factorial design, we test whether
the fairness of counter-terrorism practices increases the perceived risk of recruitment into terrorism. Second, and consequent to this contribution, we add to a
small body of legitimacy studies that apply experimental designs to test the
effects of procedural justice on people’s orientations to legal authorities (for
example, Mazerolle et al. 2013; MacQueen and Bradford 2015), thereby
responding to recurrent arguments for experimental studies of legitimacy (see,
Nagin and Telep 2017; Hyde 1983).
Unintended Negative Outcomes of Counter-Terrorism Policing: Procedural (in)Justice…
107
Procedural (in)Justice Theory and Terrorism Recruitment
Legitimacy has emerged as a central concept in criminological analyse of lawbreaking and cooperation with criminal justice agencies. It is founded on a view of human
actors as “norm-users, whose interactions with each other depend on mutually recognizable patterns that can be articulated in terms of right versus wrong conduct, or
of what one ought to do in a certain setting” (MacCormick 2007: 20). It follows that
attempts to understand people’s orientations to legal authorities demand some consideration of the legitimacy status of those authorities (Beetham 1991). The powers
of legal authorities are legitimate if they are “acknowledged as rightful by relevant
agents, who include power-holders and their staff, those subject to the power and
third parties whose support or recognition may help confirm it” (Beetham 2013:
19). A defining feature of legitimacy is that it is normative and dynamic, requiring
constant negotiation and renegotiation between relevant agents in a power relationship (Bottoms and Tankebe 2012, 2017). As Dunn (2013: 1) put it, legitimacy is “a
process of permanently active judgment (by multiple audiences), conducted with
very uneven alertness and imaginative engagement”.
There are many perspectives on legitimacy in criminology. One view measures
legitimacy in terms of citizens’ sense of obligation to obey legal authorities, institutional trust, and judgment that the values of power-holders align with those of citizens (Sunshine and Tyler 2003; Jackson et al. 2012). This is incurably deficient to
the extent that it fails to see clearly the difference between legitimacy, and trust and
obligation.1 The alternative view, as exemplified in the work of Bottoms and Tankebe
(2012, 2017), consider the multi-dimensionality of legitimacy as comprising judgments about effective exercise of authority, the actual or perceived lawfulness of
power, and fairness in distribution of resources (distributive justice) and in processes used to reach specific decisions (procedural justice) (for empirical analyses
using this latter approach, see Kearns, Ashooh and Lowrey-Kinberg 2019; Tankebe
2013; Tankebe et al. 2016). This study is situated in the Bottoms-Tankebe model,
focusing specifically on the procedural justice dimension of legitimacy.
Although dating back to Thibaut and Walker (1975), it is to Tyler that we owe the
stature of procedural justice in criminology today. A series of papers on procedure
justice culminated in Why People Obey the Law, in which Tyler (1990) outlined a
theory of procedural justice – its meaning and dimensions. Procedural Justice
describes “the fairness of the processes through which the police make decisions
and exercise authority” (Sunshine and Tyler 2003: 514). It has two interrelated
dimensions: quality of decision-making and quality of interpersonal treatment
(Sunshine and Tyler 2003). The former has three key elements: voice/participation;
neutrality; trustworthy motives. An explanation-voice dimension captures both a
normative desire for decisionmakers to given reasons for their decisions and the
This is not the place for a discussion of conceptual differences. Yet, the point must be made that
conceptual clarity is crucial to theory development and fruitful research. It is a mistake to attempt
to settle conceptual challenges by recourse to complex statistical analyses (see Blumer 1954).
1
108
J. Tankebe
extent to which they offer opportunities for civilian inputs in decision-making by
listening (attentively) to what they have to say (Tyler 1990, 2006: 116). Neutrality
hinges on people’s desire for “unbiased decision makers”, making decisions based
on pertinent objective information (Tyler 2006: 137). Thus, for example, is decision-making informed by the relevant facts or it is based on personal biases and
prejudice? The notion of trustworthy motives refers to “inferences about the intentions behind actions, intentions that flow from a person’s unobservable motivations
and character” (Tyler and Huo 2002: 61). Quality of interpersonal treatment, on the
other hand, encapsulates judgments that an individual has been treated with appropriate dignity and recognition as a person, including (where relevant) recognition of
the person’s particular needs and characteristics. Bottoms and Tankebe (2012: 145)
describe this component as the “more personal” of the elements, focused on
“whether the decision-maker treats the subject in a true sense as a human being,
with needs for dignity, privacy, respect for his or her moments of weakness,
and so on”.
Procedural justice appears empirically important because it conveys symbolic
cues about a person’s membership in society. As Tyler put it:
Although politeness and concerns for one’s rights seem like minor aspects of an interaction
with a legal authority, they convey considerable information about status within the group.
When the police harass members of minority groups, the poor, or the young, they are communicating to those groups that they have marginal social status (Tyler 2006: 176).
A number of correlational studies have shown that perceived procedural justice nurtures law-abiding behaviour and positive legal orientation (for example, Slocum and
Wiley 2018; Tankebe et al. 2016). Using data from England and Wales, Jackson and
his colleagues found evidence of positive association between procedural justice
judgements and predispositions to comply with the law (Jackson et al. 2012).
However, Augustyn (2015) reported that, among a sample of adolescent offenders,
procedural justice increased frequency of offending adolescent onset offenders.2
Similarly, Tankebe et al. (2019) reported that, on its own, procedural justice did not
correlate with commercial vehicle drivers’ self-reported violations of traffic rules;
however, an interaction effect between procedural justice and corruption experiences increased the likelihood of traffic violations.
A much smaller body of research have applied experimental designs to test the
effects of fairness treatment on orientations to legal authorities. In Queensland
(Australia), Mazerolle and her colleagues tested the effects of procedural justice
treatment on various attitudes of drivers stopped for breath tests (Mazerolle et al.
2013). They found that drivers who experienced procedurally just traffic stops were
more likely than those in a control group to express intentions to comply with traffic
laws. Drawing on data from factorial design study in the United States, Reisig et al.
(2018) tested the effects of procedural injustice by police officers in the contexts of
Augustyn argued that fair and respectful experiences might have led the adolescents to form the
view that interactions with courts and the police were not ‘that bad’ and, therefore, future interactions are not to feared.
2
Unintended Negative Outcomes of Counter-Terrorism Policing: Procedural (in)Justice…
109
traffic stops and noise complaints among a sample of undergraduate students in the
United States. They found that exposure to procedural injustice undermined the
propensity to accept police decisions and to comply with directives by officers.
Similarly, Lowrey et al. (2016) tested the effects of fair treatment by showing videos
of procedurally just and procedurally unjust traffic encounters to a sample of undergraduate students at universities in the United States. Consistent with other studies,
trust in the police was greater among those who viewed procedurally just videos
than those viewing procedurally unjust videos.
In specific area of terrorism policing, the studies have been correlational in
design. Murray, Mueller-Johnson and Sherman (2015) found that areas with high
concentrations of violent extremism were also hotspots of low confidence in the
police. Analysing data from a telephone survey of Muslim Americans in New York
City, Tyler et al. (2010) found that perceptions that police anti-terror practices were
procedural just increased people’s willingness to cooperate with the police to fight
terrorism. In Australia, Murphy, Cherney and Teston (2019) analysed data from 800
Muslims on their support for counter-terrorism policing. They found that procedural
justice perceptions increase people’s willingness to report terrorism activities to the
police. Further, feelings of social inclusion were found to predict intentions to report
terrorism. Jonathan-Zamir et al. (2016) administered a questionnaire to a sample of
1970 Arab Israeli, Jewish Israeli and foreign passengers at an airport in Israel to test
the influence of procedural justice on reactions to enhanced security screening tactics. They found that experiencing the screening as being procedurally just decreased
the feeling of humiliation, intimidation and insensitivity. Finally, Doosje et al.
(2013) surveyed 131 Dutch Muslim youths to investigate their views on violent
extremism by fellow Muslims and their own violent intentions. They found that
perceived procedural injustice weakened the bonds between the youths and Dutch
society, which, in turn, increased their acceptance of violence by fellow Muslims.
In summary, the procedural justice component of legitimate power places emphasis on the perceived morality of processes in the use of authority. The key moral
ingredients are participatory decision-making, respectful and dignified treatment, neutrality, and the trustworthy motives of power-holders. These moral judgements have been shown to induce a feeling of obligation to obey rules and
institutions, which, in turn, leads to actual and intended behaviours across different
settings. However, while the research evidence predominantly supports the theoretical arguments, the literature is not without limitations. Two of these are pertinent here.
The first concerns evidence of contextual influences such that procedural justice
offers “no simple recipe for winning legitimacy” (Waddington et al. 2015: 212; see
also, Worden and McLean 2017). Waddington and his colleagues found, through
analyses of public evaluations of videos of police behaviour, that a history of difficult police-public relations can mean people enter interactions with priors that
sometimes neutralise the effects of procedural justice. Tyler (2000) also showed
that, when people have strong bonds with subgroups, procedural justice tended to be
a secondary normative expectation when they interacted with power-holders.
Second, there are concerns about the evidence base for procedural justice theory,
110
J. Tankebe
especially as this relates to crime prevention. Thus, in their review of the evidence,
Weisburd and Majmundar (2018: 8) conclude that:
There is currently only a very small evidence base from which to support conclusions
about the impact of procedural justice policing on crime prevention. Existing research
does not support a conclusion that procedural justice policing impacts crime or disorder
outcomes. At the same time, because the evidence base is small, the [Committee on
Proactive Policing] also cannot conclude that such strategies are ineffective (see also
Nagin and Telep 2017).
Consequently, the evidence requires us to adopt an agnostic position on the effects
of procedural justice on criminal involvement. The two limitations are not mutually
exclusive: contextual and relational variations add to the challenges of building
more robust evidence base for procedural justice. Additionally, the predominance of
correlational studies over those using experimental designs makes it difficult to
establish causal links. In using a factorial experimental design to test procedural
justice effects among a national sample of Muslims, the current study represents a
major step in addressing some of the limitations in the literature.
Terrorism and Counter-Terrorism in the United Kingdom
Terrorism, in its various forms, have a long history in the United Kingdom. However,
the most recent history dates to the London underground bombings of July 2005
which killed 52 people. These were followed by other attempted and successful
attacks. For example, in May 2017, a young man detonated homemade bomb as
young people made their way out of musical concert in Manchester, killing 22 of
them. It was one of five attacks in 2017, including those at Westminster Bridge and
the Houses of Parliament in London, resulting in 36 deaths. According to official
statistics, there were 2029 terrorism arrests between 2010 and 2017 in Great Britain,
and a further 500 ongoing investigations, involving 3000 individuals (Home
Office 2018).
The counter-terrorism strategy for the United Kingdom, known as CONTEST,
has four strands: prevent (“stop people from becoming terrorists or supporting
terrorism”); pursue (stop terrorist); protect (“strengthen … protection against a
terrorist attack”) and prepare (“mitigate the impact of a terrorist attack”) (Home
2018: 8). These strands translate into various everyday policing tactics such as
stop and searches for the purposes of terrorism prevention. The most notable
counter-­terrorism legislations include Terrorism Act 2000, Terrorism Act 2006,
and Counter-­terrorism and Security Act 2015. Section 43 of Terrorism Act 2000
grants police constables the power to stop and search a person whom they “reasonably suspects to be a terrorist to discover whether he has in his possession
anything which may constitute evidence that he is a terrorist.” Between June 2017
and June 2018, the London Metropolitan police stopped and searched 644 people
under this legislation. For the whole of 2017, 767 searches were conducted in
London under s.43.
Unintended Negative Outcomes of Counter-Terrorism Policing: Procedural (in)Justice…
111
Whether they are for the purposes of tackling terrorism or other offences, stop
and search powers allow officers to “allay or confirm suspicions about individuals
without exercising their power of arrest” (PACE Code of Ethics, 2015). The use of
these statutory powers is governed by a code of practice. It entreats officers to act
with fairness and respect towards persons subject these powers, and to eschew
harassment and discrimination. However, there is often a gap between the law and
practice. According to Bowling and Phillips (2007):
The power to stop and search is an investigative power used for the purposes of crime detection or prevention in relation to an individual suspected of a specific offence at a specific
time. In practice however, police officers frequently use stop and search powers for other
purposes such as ‘gaining intelligence’ on people who are ‘known’ to the police, to break
up and move on groups of people, and for the purposes of ‘social control’ more generally.
Although there is no basis in law for the police to use the power to stop and search for these
purposes, the practice is widespread (p. 938).
Research evidence shows that most of police stop and search practices produce false
positives. Ariel and Tankebe (2018) analysed five-year data from a police force in
England and found that 66 percent of police stop and search activities resulted in no
action taken against those individuals – no warnings, formal advice, or arrests. Of
the 767 s.43 searches conducted in London in 2017, only 8 percent resulted in
arrests (Hill 2018). However, false positives are only one of many concerns about
the use of stop and search powers in the UK: for example, there are concerns that
the police disproportionately target minorities and treat them in ways that are disrespectful and undignified. Thus, on the former, Hargreaves (2018) analysed stop and
search data between 2006 and 2011 and found that pedestrians who self-identified
as Muslim were at a slightly higher risk of being stopped. He further found that such
stops for Muslims were eight times more likely than those for any other ethnic or
religious groups to result in searches.
Loader (2006: 211) has put it thus:
Every stop, every search, every arrest, every group of youths moved on, every abuse of due
process, every failure to respond to call or complaint, every racist snub, every sexist remark,
every homophobic joke, every diagnosis of the crime problem, every depiction of criminals – all these send small, routine, authoritative signals about society’s conflicts, cleavages,
and hierarchies, about whose claims are considered legitimate within it, about whose status
identity is to be affirmed or denied as part of it (emphasis added).
Loader’s claims chimes with Tyler’s (2011) assertion that routine interactions
between police officers and civilians are teachable moments that might contribute to
nurturing or squandering legitimacy. In their analysis of survey data from young
people about their stop and search experiences in New York City, Tyler et al. (2014)
found that perceived unfairness of these proactive contacts undermined attitudes
towards the police. In the UK, Bradford (2017) finds that police stops perceived as
unfair undermine overall legitimacy assessments of the police. Thus, what emerges
from this brief excursus is the fraught nature of police stop and search powers, and
the implications for people’s reactions to the law. What remains unexamined is how
far exposure to procedurally (un)just stops affect perceived risk of terrorism recruitment. The results presented below offer some insight.
112
J. Tankebe
Data and Methods
To test the hypothesis that procedural injustice affects perceived risk of recruitment
into terrorism, we needed an experiment that revealed the causal relationship.
Factorial surveys allowed us to accomplish that objective. They involve vignettes
that provide “short descriptions of a person or a social situation which contain precise references to what are thought to be the important factors in the decision-­
making or judgment-making processes of respondents” (Alexander and Becker
1978: 94). Factorial surveys are particularly useful in situations in which there are
challenges to conducting actual experimental manipulations. It was precisely the
situation we faced: it was impossible to obtain police approval randomly to assign
officers in procedural justice scripts for the purposes of testing their effects on risk
of recruitment. Although, factorial designs have their weakness (see Wallander
(2009) for an overview of these designs), they have “the ability to establish causal
inference and isolate underlying mechanisms while using realistic contexts that
alleviate potential concerns about external validity” (Schilke and Rossman 2018: 10).
The Procedure
The study took place between February and March 2018. We contracted a private
research company, Survation, to conduct a nationwide telephone survey of adult (18
and older) Muslims in the United Kingdom. The sampling process involved selecting small geographic units called ‘lower super output areas’ (LSOA) in which people self-identifying as Muslims constituted at least 20 percent of the residents. This
information came from the 2010 national census data. All postcodes belonging to
the selected LSOAs were extracted from a database of active residential postcodes
and the contact details of the residents were retrieved. Survation has developed a
strong ability for targeting through telephone specific individuals according to their
socio-demographic characteristics, increasing the possibility of generating a representative sample of the population. In total, 1019 Muslims responded to researchers’ phone calls of which 798 agreed to participate in the study, representing a
take-up rate of 78.3 percent. Of the 221 who declined involvement, 93 had originally consented but dropped out later in the study; the rest did not start at all.
The researchers explained the purpose of the study and that it was entirely voluntary and anonymous. Each participant was required to indicate expressly whether
they understood the information provided and were happy to consent to take part in
the study; if they declined the researchers thanked them and the call ended. If they
agreed to take part in the study, their consent was recorded, and the interview started
with initial questions on age, gender, educational attainment and employment status. The participants were not offered any incentives for participating in the study.
The study involved four scenarios and each research participant was randomly
assigned to receive only one. The first scenario depicted a fictitious street interaction
Unintended Negative Outcomes of Counter-Terrorism Policing: Procedural (in)Justice…
113
between a young man and police officers in which the officer acted procedurally
justly (high procedural justice condition). It read as follows:
A young Muslim man has recently been released from prison three years ago. He has since been
stopped and searched several times. In recent months, the police demand additional information
from him and they do so in ways that he finds respectful. The young man feels the officers show
care for his well-being and always listen carefully to what he has to say for himself.
The second scenarios also involved a street stop but low in procedural justice:
A young Muslim man has recently been released from prison three years ago. He has since
been stopped and searched several times. In recent weeks, the police demand additional
information from him and do so in a way he finds disrespectful. He feels the officers show
no care for his well-being and they do not take time to listen to what he has to say for
himself.
The third and fourth scenarios described police raids at a Muslim family in which a
young man was arrested on suspicion of terrorism. Given the nature of counter-­
terrorism raids, the experimental treatment focused on police actions after the raid.
The high procedural justice raid condition read as follows:
A family home of one of your friends is raided at night by a special counter-terrorism unit
of police officers, which involved the breaking of the front door. The police entered in their
riot gear, guns and masks. This caused disruption and fear in the family who felt violated by
the police. A young man was arrested but later released without a charge. A police officer
came to the house the next morning to explain the reasons why the officers had carried out
the night time raid, and to listen to the concerns of the family.
In the low procedural justice condition, the research participants were read the following scenario:
The family home of one of your friends is raided at night by a special counter-terrorism unit
of police officers, which involved the breaking of the front door. The police entered in their
riot gear, guns and masks. This caused disruption and fear in the family who felt violated by
the police. A young man was arrested but later released without a charge.
After reading out each scenario, participants were each asked questions that measured perceived risk of recruitment into terrorism. The questions were: (1) “How
likely is it that this young man will be attracted to beliefs that justify the use of
violence to bring change in society?”; (2) “How likely is it that this young man will
engage in violent activity as a way to bring change in society?”; and (3) “How likely
is it that this young man’s experience will lead him to support violent activity as a
way to bring change in society?” the response options ranged from 0 = Not likely at
all to 3 = Extremely likely. The Cronbach alpha reliability score was 0.68 for the
police stops and 0.61 for police raids.
Basic Sample Characteristics
The age of the participants in the procedurally just police stop ranged from 18 to
74 years (M = 35.09) while procedural injustice stops ranged from 18 to 72 years
(M = 33.71). In the case of counter-terrorism raid, the ages of the participants ranged
114
J. Tankebe
from 18 to 85 years for those in procedurally just condition (M = 36.9 years) 18 to
81 years for those in the procedural unjust conditions (M = 36.5 years). Gender is
coded as male = 1 and female = 2. Education is dichotomised into pre-university
education = 1 and university-level education = 2. Personal contacts with police was
coded as ‘no contacts’ = 1 and ‘any contacts’ = 2. A description of the sample is
presented in Table 1.
Results
Independent t-tests were used to compare the mean differences between conditions
of procedural injustice and procedural justice in perceived risk of terrorism recruitment. The results are presented in Table 2. Starting with the police stops scenarios,
we found that perceived risks of recruitment into terrorism was greater for those
exposed to procedurally unjust stops (M = 1.38, SD = 0.58) than those exposed
Table 1 Summary Statistics for Each Experimental Condition
Street Stops
Police Raids
High PJ
Low PJ
High PJ
Age (Mean/SD)
35.1 (SD = 13.9) 33.7 (SD = 12.7) 36.9
(SD = 14.7)
Sex = female
50.8
56.3
52.7
Education = University 39.3
44.8
38.4
Personal contact = Yes 36.8
39.3
45.9
Employment
Full-time
37.6
47.1
37.3
Part-time
18.1
17.5
16.5
Others
44.3
35.5
46.1
Low PJ
36.5 (SD = 14.3)
44.0
44.6
44.1
48.5
12.0
39.5
Table 2 T-tests for Perceived Risk of Terrorism Recruitment
Police Stops
Procedural Procedural
Justice
Injustice
Mean
Mean
(SD)
(SD)
Risk of
1.21
1.38
recruitment (0.53)
(0.58)
Terrorism 1.32
1.38
attraction (0.77)
(0.76)
Terrorism 1.14
1.31
activity
(0.67)
(0.66)
Terrorism 1.17
1.46
support
(0.63)
(0.75)
n
t
df
330 −2.81
328
330 −0.67
328
330 −2.22∗ 328
330 −3.85
∗∗∗
Note. ∗p < .05, ∗∗p < .01, ∗∗∗p < .001
320
Police Raids
Procedural Procedural
Justice
Injustice
Mean
Mean
(SD)
(SD)
1.28
1.33
(0.55)
(0.53)
1.28
1.33
(0.75)
(0.71)
1.29
1.29
(0.72)
(0.771)
1.27
1.37
(0.73)
(0.69)
n
t
df
301 −0.82 299
301 −0.71 299
301 0.04
299
301 0.76
299
Unintended Negative Outcomes of Counter-Terrorism Policing: Procedural (in)Justice…
115
to procedurally just police stops (M = 1.21, SD = 0.53). The magnitude of the difference between the two participants was small, Cohen’s d = −0.31. We disaggregated
the measure of recruitment risk into three specific indicators: attraction to terrorism;
engaging in terrorism activity; and support for terrorism. Perceived risk of attraction
to terrorism did not differ between the procedurally unjust and procedurally just
conditions. However, the research participants exposed to procedurally unjust police
stops perceived a greater risk of the young man engaging in terrorism activity
(Cohen’s d = −0.26) and supporting terrorism (Cohen’s d = −0.42) than those
exposed to procedurally just police stops. Next, we consider the scenarios involving
police counter-terrorism raids at a home. Overall, we found no statistically significant difference in perceptions of the risk of recruitment into terrorism between those
exposed to high procedural justice condition (M = 1.33, SD = 0.53) and those
exposed to low procedural justice raids (M = 1.28, SD = 0.55). Nor did any mean
differences emerge around the individual disaggregated indicators of perceived
recruitment risk.
Discussion and Conclusion
Terrorism is a major security issue for many countries. Consequently, “[u]nderstanding, predicting, and thwarting extreme behavior, particularly as it relates to
violence, is one of the most challenging issues facing humankind” (Gelfand et al.
2013: 495). However, among the key issues in counter-terrorism are the risks of
violent radicalisation and recruitment into terrorism and the extent to which counter-­
terrorism practices might be doing more harm than good. Specifically, there have
been concerns that the nature of counter-terrorism can have unintended negative
effects, increasing rather than reducing the risks of recruitment into terrorism.
Do counter-terrorism practices have unintended negative effects? Using a
nationally-­representative sample of Muslims in the United Kingdom, this chapter
has reported findings from a factorial design that tested the effects of procedural (in)
justice of counter-terrorism practices on perceived risks of recruitment into terrorism. The study offers interesting results with important implications for terrorism
researchers, policy mangers, and counter-terrorism officials. First, we used descriptive analysis to consider the perceived risk of recruitment into terrorism under conditions of procedurally (un)just conditions. We found small and statistically
significant difference in perceived recruitment risk between two conditions relating
to police stops, such that those exposed to procedurally unjust police stops perceived a greater risk of terrorism recruitment than those exposure to procedurally
just stops.
This finding that procedurally unjust street-level encounters are perceived to
pose damaging effect on terrorism prevention would seem consistent with evidence
from procedural justice theory. That evidence shows that poor treatment undermines
people's actual or intended orientations to the law. In Queensland (Australia), police
officers were randomly assigned to receive ‘procedural justice script’ to guide their
116
J. Tankebe
routine interactions with motorists. Survey data from motorists in both control and
experimental conditions were analyzed to test the effects of procedural justice perceptions on people’s orientations towards the police. However, as Jonathan-Zamir,
Hasisi and Margalioth (2016: 632) concluded from their study of passengers subjected to enhanced screening at an airport in Israel: “it is not the case that they can
engage in any intrusive or unpleasant practice, and be perceived positively, just as
long as they follow the principles of procedural justice. Some policing tactics,
which we suspect are particularly harassing or suggest inequity in their distribution,
will likely produce negative emotions even if they are executed in a procedurally-­
just fashion.” Consequently, the results presented above cannot be interpreted as
evidence that the quality of experiences with security agencies play a decisive role
in perceptions of the risk of unintended effects of terrorism recruitment. Indeed, the
differences between our treatment groups in perceptions of risks of recruitment
were small.
Procedural justice theory envisages the application of its principles in the course
of specific decision-making or the general exercise of power. However, in counter-­
terrorism contexts, this expectation cannot be taken for granted. It sometimes happens that counter-terrorism raids occur in highly volatile and potentially dangerous
contexts, leaving little opportunity for the in-situ application of procedural justice
principles. This raises an important question that had hitherto been unaddressed:
what effects do post-operational actions mimicking procedural justice principles
have on the perceived risk of unintended negative consequences of counter-­terrorism
practices? We investigated this question and found no evidence that procedurally
unjust post-raids affected perceived risks of terrorism recruitment.
These findings suggest two things: first, potential differences in procedural justice effects across counter-terrorism practices (street stops versus domestic raids).
Second, the findings might be taken to imply that procedural justice makes no difference to public attitudes if it is a post-facto strategy of community engagement. It
neither attenuates nor worsens the effects of counter-terrorism practices. It is well
known that absence of evidence of effects is not evidence that the effects are absent
(see, Taleb 2007). It sometimes happens that there are methodological limitations
that prevent researchers from detecting effects. In the current study, it is possible
that our design of the vignettes contained weaknesses. Part of the challenge is that
counter-terrorism raids are a rare event, hence a difficulty in finding a sample with
such experiences for a pilot study that could have improved our vignette. Future
studies that address this limitation might advance our knowledge of the effects of
procedural justice beyond the relatively mundane contests in which it has been
tested in much of the extant literature.
To conclude, we have shown that the self-identified Muslims surveyed in the
United Kingdom believed that procedurally unjust police stops increased the likelihood of recruitment into terrorism. To be sure, we measured the participants perceptions of the vulnerability of others rather than themselves. However, given the social
psychological evidence that people tend to project their views onto others (for
example, Kawada et al. 2004), the risk assessments might be indicative of the participants’ own propensities, and therefore what they might be tempted to do if sub-
Unintended Negative Outcomes of Counter-Terrorism Policing: Procedural (in)Justice…
117
jected to counter-terrorism practices that they perceive to be unjust. Finally,
procedure is only one dimension of justice perceptions. Justice also has a substantive dimension, a point that is often underemphasised in the procedural justice literature (For an exception, see Peffley and Hurwitz 2010; Tankebe 2013; Bottoms
and Tankebe 2017; McLean 2019). Critical legal theorists warn that “an appearance
of justice is often fabricated, usually by those in power, and often through deceit”
such that “the offer to ‘hear you out,’ to ‘invite your participation,’ to ‘listen to what
you have to say (for yourself)’” sometimes distracts us from injustices in society
(Cohen 1989: 32). Counter-terrorism policing is not immune to that caution.
Acknowledgement This research was supported by funding from the European Union’s Horizon
2020 research and innovation programme under grant agreement #699824.
References
Alexander, C. S., & Becker, H. J. (1978). The use of vignettes in survey research. Public opinion
Quarterly, 42(1), 93–104.
Annan, K. (2003). Ability to reason vital in fighting terrorism, Secretary-General Tells Conference.
Available at https://www.un.org/press/en/2003/sgsm8885.doc.htm Accessed 22 September
2018.
Ariel, B., & Tankebe, J. (2018). Racial stratification and multiple outcomes in police stops and
searches. Policing and Society, 28(5), 507–525.
Augustyn, M. B. (2015). The (ir)relevance of procedural justice in the pathways to crime. Law and
Human Behavior, 39(4), 388–401.
Beetham, D. (1991). The legitimation of power (1st ed.). London: Palgrave Macmillan.
Beetham, D. (2013). The legitimation of power (2nd ed.). Basingstoke: Palgrave Macmillan.
Blumer, H. (1954). What is wrong with social theory? American Sociological Review, 19(1), 3–10.
Borum, R. (2011). Radicalization into violent extremism I: A review of social science theories.
Journal of Strategic Security, 4(4), 7–36.
Bottoms, A. E., & Tankebe, J. (2012). Beyond procedural justice: a dialogic approach to legitimacy
in criminal justice. Journal of Criminal Law and Criminology, 102,119–70.
Bottoms, A., & Tankebe, J. (2017). Police legitimacy and the Authority of the State. In M. Ulväng,
A. du Bois-Pedain, & P. Asp (Eds.), Criminal Law and the Authority of the State (pp. 47–88).
Oxford: Hart Publishing.
Bowling, B., & Phillips, C. (2007). Disproportionate and discriminatory: Reviewing the evidence
on police stop and search. The Modern Law Review, 70(6): 936–961.
Bradford, B. (2017). Stop and search and police legitimacy. London: Routledge.
Choudhury, T., & Fenwick, H. (2011). The impact of counter-terrorism measures on Muslim communities. International Review of Law, Computers & Technology, 25(3), 151–181.
Cohen, R. L. (1989). Fabrications of justice. Social Justice Research, 3, 31–46.
Cronin, A. K. (2009). How terrorism ends: Understanding the decline and demise of terrorist
campaigns. Princeton: Princeton University Press.
Dalgaard-Nielsen, A. (2010). Violent radicalization in Europe: What we know and what we do not
know. Studies in Conflict & Terrorism, 33(9), 797–814.
Della Porta, D. (2018). Radicalization: A relational perspective. Annual Review of Political
Science, 21(1), 461–474.
Doosje, B., Loseman, A., & Van Den Bos, K. (2013). Determinants of radicalization of Islamic
youth in the Netherlands: Personal uncertainty, perceived injustice, and perceived group threat.
Journal of Social Issues, 69(3), 586–604.
118
J. Tankebe
Dunn, J. (2013). Legitimacy and democracy in the world today. In J Tankebe & A Liebling (eds.),
Legitimacy and criminal justice: An international exploration,Oxford: Oxford University Press.
Gelfand, M. J., LaFree, G., Fahey, S., & Feinberg, E. (2013). Culture and extremism. Journal of
Social Issues, 69(3), 495–517.
Hargreaves, J. (2018). Police stop and search within british muslim communities: Evidence from
the crime survey 2006–11. British Journal of Criminology, 58(6), 1281–1302.
Hill, M. (2018). The Terrorism Acts in 2017. Available at https://terrorismlegislationreviewer.independent.gov.uk/wp-content/uploads/2018/10/The_Terrorism_Acts_in_2017.pdf [Accessed 22
November 2018].
Home Office. (2018). CONTEST: The United Kingdom’s Strategy for Countering Terrorism (Cm
9608): Annual Report. London: The Stationery Office.
Hyde, A. (1983). The concept of legitimation in the sociology of law. Wisconsin Law Review, 1983,
379–426.
Jackson, J., Bradford, B., Stanko, B., & Hohl, K. (2012). Just authority?: Trust in the police in
England and Wales. Oxford: Routledge.
Jonathan-Zamir, T., Hasisi, B., & Margalioth, Y. (2016). Is it the what or the how? The roles of
high-policing tactics and procedural justice in predicting perceptions of hostile treatment: The
case of security checks at Ben-Gurion Airport, Israel. Law & society review, 50(3), 608–636.
Kawada, C. L., Oettingen, G., Gollwitzer, P. M., & Bargh, J. A. (2004). The projection of implicit
and explicit goals. Journal of Personality and Social Psychology, 86(4), 545–559.
Kearns, E. M., Ashooh, E., & Lowrey-Kinberg, B. (2019). Racial differences in conceptualizing
legitimacy and trust in police. American Journal of Criminal Justice, 1–25.
LaFree, G., Jensen, M. A., James, P. A., & Safer-Lichteinstein, A. (2018). Correlates of violent
political extremism in the United States. Criminology, 56(2), 233–268.
Loader, I. (2006). Policing, recognition, and belonging. The Annals of the American Academy of
Political and Social Science, 605(1), 201–221.
Lowrey, B. V., Maguire, E. R., & Bennett, R. R. (2016). Testing the effects of procedural justice and overaccommodation in traffic stops: A randomized experiment. Criminal justice and
Behavior, 43(10): 1430–1449.
MacCormick, N. (2007). Institutions of law. Oxford: Oxford University Press.
MacQueen, S., & Bradford, B. (2015). Enhancing public trust and police legitimacy during
road traffic encounters: Results from a randomised controlled trial in Scotland. Journal of
Experimental Criminology, 11(3), 419–443.
Mazerolle, L., Antrobus, E., Bennett, S., & Tyler, T. R. (2013). Shaping citizen perceptions of
police legitimacy: A randomized field trial of procedural justice. Criminology, 51, 33–63.
McCauley, l., & Moskalenko, S. (2008). Mechanisms of political radicalization: Pathways towards
terrorism. Terrorism and Political Violence, 20(3), 415–433.
McLean, K. (2019). Revisiting the role of distributive justice in Tyler’s legitimacy theory. Journal
of Experimental Criminology. https://doi.org/10.1007/s11292-019-09370-5.
Mumpower, J. L., Shi, L., Stoutenborough, J. W., & Vedlitz, A. (2013). Psychometric and demographic predictors of the perceived risk of terrorist threats and the willingness to pay for terrorism risk management programs. Risk Analysis, 33(10), 1802–1811.
Murphy, K., Cherney, A., & Teston, M. (2019). Promoting Muslims’ willingness to report terror
threats to police: testing competing theories of procedural justice. Justice Quarterly, 36(4),
594–619.
Murphy, K., Madon, N. S., & Cherney, A. (2017). Promoting Muslims’ cooperation with police
in counter-terrorism: The interaction between procedural justice, police legitimacy and law
legitimacy. Policing: An International Journal, 40(3), 44–559.
Murray, A., Mueller-Johnson, K., & Sherman, L. W. (2015). Evidence-based policing of UK
Muslim communities: linking confidence in the police with area vulnerability to violent
extremism. International criminal justice review 25(1), 64–79.
Nagin, D. S., & Telep, C. W. (2017). Procedural justice and legal compliance. Annual Review of
Law and Social Science, 13, 5–28.
Unintended Negative Outcomes of Counter-Terrorism Policing: Procedural (in)Justice…
119
Pantazis, C., & Pemberton, S. (2009). From the ‘Old’ to the ‘New’ suspect community examining the impacts of recent UK counter-terrorist legislation. The British Journal of Criminology,
49(5), 646–666.
Peffley, M., & Hurwitz, J. (2010). Justice in America: The separate realities for blacks and whites.
Cambridge: Cambridge University Press.
Reisig, M. D., Mays, R. D., & Telep, C. W. (2018). The effects of procedural injustice during
police–citizen encounters: A factorial vignette study. Journal of Experimental Criminology,
14(1), 49–58.
Schilke, O., & Rossman, G. (2018). It’s only wrong if it’s transactional: Moral perceptions of
obfuscated exchange. American Sociological Review, 83, 1079–1107.
Slocum, L. A., & Wiley, S. A. (2018). “Experience of the expected?” Race and ethnicity differences in the effects of police contact on youth. Criminology, 56(2), 402–432.
Sunshine, J., & Tyler, T. R. (2003). The role of procedural justice and legitimacy in shaping public
support for policing. Law & Society Review, 37, 513–48.
Taleb, N. N. (2007). The Black Swan: The Impact of the Highly Improbable. New York: Random
House.
Tankebe, J. (2013). Viewing things differently: The dimensions of public perceptions of police
legitimacy. Criminology, 51(1), 103–135.
Tankebe, J., Reisig, M. D., & Wang, X. (2016). A multidimensional model of police legitimacy: A
cross-cultural assessment. Law and Human Behavior, 40(1), 11–22.
Tankebe, J., Boakye, K. E., & Amagnya, A. M. (2019). Traffic violations and cooperative intentions among drivers: The role of corruption and fairness. Policing and Society, 1–16. https://
doi.org/10.1080/10439463.2019.1636795.
Tyler, T. R. (1990). Why people obey the law (1st ed.). New Haven: Yale University Press.
Tyler, T. R. (2000). Multiculturalism and the Willingness of Citizens to Defer to Law and to Legal
Authorities. Law & Social Inquiry, 25(4): 983-1019
Tyler, T. R. (2006). Why people obey the law (2nd ed.). Princeton: Princeton University Press.
Tyler, T. R. (2011). Trust and legitimacy: Policing in the USA and Europe. European Journal of
Criminology, 8, 254–266.
Tyler, T. R., & Huo, Y. (2002). Trust in the law: Encouraging public cooperation with the police
and courts. New York: Sage.
Tyler, T. R., Schulhofer, S., & Huq, A. Z. (2010). Legitimacy and deterrence effects in counterterrorism policing: A study of Muslim Americans. Law & Society Review, 44(2), 365–402.
Tyler, T. R., Fagan, J., & Geller, A. (2014). Street stops and police legitimacy: Teachable moments
in young urban men’s legal socialization. Journal of Empirical Legal Studies, 11(4), 751–785.
Waddington, P. J., Williams, K., Wright, M., & Newburn, T. (2015). Dissension in public evaluations of the police. Policing and Society, 25, 212–235.
Wallander, L. (2009). 25 years of factorial surveys in sociology: A review. Social Science Research,
38(3), 505–520.
Weisburd, D., & Majmundar, M. K. (2018). Proactive Policing: Effects on Crime and Communities.
Washington, DC: The National Academies Press.
Worden, R. E., & McLean, S. J. (2017). Mirage of police reform: Procedural justice and police
legitimacy. Oakland, CA: University of California Press.
The Link Between Prior Criminal Record
and Violent Political Extremism
in the United States
Michael A. Jensen, Aaron Safer-Lichtenstein, Patrick A. James,
and Gary LaFree
Introduction
Research on radicalization1 has progressed a great deal over the past decade, as
scholars have incorporated views from economics, political science, psychology,
and sociology into increasingly comprehensive explanations of violent extremism.
While criminology has been somewhat slower than some of the other social sciences to fully engage in the study of radicalization and violent extremism, the situation is changing rapidly (Forst et al. 2011; LaFree and Freilich 2017). Research on
political extremism now routinely appears in major criminology outlets, criminology students are defending theses and dissertations on related topics, and meetings
of professional criminology associations have a strong representation of terrorism-­
related research. Indeed, recent efforts (Jensen et al. 2016; LaFree et al. 2018) to
bring criminological insights into the study of political extremism have found that
common criminological correlates explain a good deal of extremist behavior.
We adopt the FBI definition of radicalization as “the process by which individuals come to believe
their engagement in or facilitation of nonstate violence to achieve social and political change is
necessary and justified” (Hunter and Heinke 2011).
1
This research was supported by the European Union’s Horizon 2020 research and innovation program under award number 699824, the National Institute of Justice through award number 2012-­
ZA-­BX-0005, and the Department of Homeland Security Science and Technology Directorate’s
Office of University Programs through Award Number 2012-ST-061-CS0003. The views and conclusions contained in this document are those of the authors and should not be interpreted as necessarily representing the official policies, either expressed or implied, of the European Union, the
U.S. Department of Justice, or the U.S. Department of Homeland Security.
M. A. Jensen (*) · A. Safer-Lichtenstein · P. A. James · G. LaFree
University of Maryland, College Park, MD, USA
e-mail: majensen@umd.edu
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_6
121
122
M. A. Jensen et al.
At the heart of these efforts is the question of how similar terrorism-related crime is
to more ordinary types of crime. Criminologists Clarke and Newman (2006) see the
overlap as complete, claiming that “terrorism is a form of crime in all essential respects.”
While LaFree and Dugan (2015) substantially agree with this conclusion, they point
out that there are nonetheless important differences between terrorism-­related crime
and more ordinary types. Perhaps most notably, in contrast to many who engage in terrorist activity, ordinary criminals rarely see their illegal actions as altruistic.
Nonetheless, it seems clear that terrorist behavior shares much with ordinary
crime. Of particular importance for the future of terrorism research are the recent
findings (Jensen et al. 2016, 2018; LaFree et al. 2018) that previous non-ideological
criminality—a correlate that is included in most etiological studies of crime—may
help distinguish extremists who participate in acts of politically motivated violence
from those who commit illegal actions driven by political motives, but do not reach
the threshold of violence. For example, several studies of terrorism in the United
States (Gruenewald et al. 2013; Jensen et al. 2016, 2018; LaFree et al. 2018) have
found that extremists are more likely to engage in violence in pursuit of their political ambitions if they have a history of engaging in criminal acts prior to radicalizing.
Similar results have been found in the radicalization processes of European jihadists
(Basra and Neumann 2016).
Taken as a whole, the evidence suggests that criminological perspectives emphasizing the link between previous criminality and future criminality is important for
understanding political extremism. While radicalization experts have begun to
explore these links, several important questions about the relationship between previous criminal behavior and extremist violence remain. For example, very little is
known about how frequently extremists engage in non-ideological crimes prior to
radicalizing or which crimes they are most likely to commit. Similarly, the field of
terrorism studies has yet to address whether juvenile offending (in contrast to adult
offending) changes the risk calculus for violence among political extremists. This
chapter seeks to build on the emerging body of research that explores the relationship between established correlates of crime developed in the criminology literature
and the etiology of violent extremism by addressing these questions.
Using data from the newly released Profiles of Individual Radicalization in the
United States (PIRUS) database, we explore the criminal histories and extremist
behaviors of nearly 1900 individuals who radicalized in the United States since
1948. Our results reveal that pre-radicalization criminal behavior, violent or nonviolent, is the single strongest non-ideological predictor of post-radicalization violence among U.S. extremists. However, we find that the criminal backgrounds of
U.S. extremists vary considerably depending on individuals’ ideological affiliations. Individuals on the extremist far-right, especially those motivated by white
supremacist views, are substantially more likely to engage in crime prior to radicalizing than are individuals associated with other ideologies. We also find that
U.S. extremists rarely specialize in specific crime types prior to or after radicalizing. Rather, individuals in the PIRUS data with criminal histories are often repeat
offenders who engaged in violent and non-violent crimes or non-violent criminals
who escalated to violent crime once they radicalized. Finally, we find that while the
The Link Between Prior Criminal Record and Violent Political Extremism in the United… 123
rates of juvenile offending among U.S. extremists are not substantially higher than
they are for the general population, individuals who engage in criminal activity
before the age of 18 are significantly more likely than non-juvenile offenders to
engage in acts of violent extremism after radicalizing.
This chapter proceeds in five sections. First, we review the criminological literature on the link between past and future crime, as well as recent attempts by terrorism scholars to bring criminological theory into the study of extremists. Second, we
describe the PIRUS data and the additional data collection that we conducted to
explore the link between previous criminality and violent extremism. Third, we
present descriptive findings on the past criminal behaviors of U.S. extremists,
including how crime rates compare across ideologies and whether juvenile crime is
present at higher rates in the backgrounds of violent extremists. Fourth, we present
a multivariate analysis that tests the relationships between a number of ideological
and non-ideological correlates and violent extremism, including the effect that pre-­
radicalization crime has on the odds that an extremist will engage in violence after
radicalizing. Finally, we conclude by considering the implications of this research
for terrorism prevention policy and future studies of terrorism.
Literature Review
While terrorism studies have begun to establish links between non-ideological
criminal behavior and extremism, important research gaps remain. Criminological
investigations of ordinary crime often emphasize the role that previous criminality,
crime diversity, and juvenile offending play in the development of criminal trajectories. In particular, these factors may be important in an offender’s evolution towards
increasingly extreme forms of delinquency, including acts of criminal violence.
Below, we review the criminological research on these issues and assess the extent
to which they have been addressed by terrorism researchers.
Previous Criminal Behavior
Decades of criminological research has consistently found a robust relationship
between prior deviance and subsequent criminal behavior (Blumstein et al. 1988;
Loeber and Le Blanc 1990). Indeed, most actuarial assessments of sentencing,
parole, and probation decisions rely on prior criminal record (Hoffman and Beck
1974; Hoffman et al. 1978; Monahan and Skeem 2015). In an influential meta-­
analysis of the predictors of adult offender recidivism, Gendreau et al. (1996) examined 131 prior studies and found criminal history to be among the strongest
predictors of future criminal activity. Similarly, DeLisi et al. (2013) found that prior
police contact and arrest were predictive of future violent crime.
But while the association between criminal record and ordinary crime is well
established, there is far less research on the connection between criminal record and
124
M. A. Jensen et al.
political violence (Cottee 2016).2 Only recently, a few studies have included criminological variables in analyses predicting violent extremism. For example,
Gruenewald et al. (2013) examined the correlates of far-right loner and far-right
group-motivated homicides, including the impact of criminogenic factors. The
authors found that 29 out of 47 (61.7%) loners and 47 out of 92 (51.1%) group-­
motivated extremists had prior arrests; 16 out of 29 loners (55.2%) and 21 out of 47
(44.7%) group-motivated offenders had violent criminal histories. Similarly, a
recent LaFree et al. (2018) study examined correlates of violent extremism in a
sample of U.S. extremists and found that 40% of the subjects engaged in crime prior
to adopting extremist beliefs. These rates exceed most contemporary estimates of
cumulative arrest histories in the general population (Brame et al. 2014). While
there is no official measure of crime prevalence in the U.S., Brame et al. (2014)
estimate that 25–41% of young adults in the U.S. have an arrest history (though
subsequent research has found that these rates vary significantly across measures of
gender and race) and the Federal Bureau of Investigation maintains felony arrest
records for approximately one third of the U.S. population (Clark 2017). Substantial
evidence of prior criminal records among political extremists has been reported by
national police and intelligence agencies outside of the U.S. as well. For example,
officials from Germany, Norway, and the Netherlands reported that more than 60%
of the individuals that traveled from their countries to join the Islamic State in Syria
and Iraq had previous criminal records (Basra and Neumann 2016).
Importantly, recent studies (Jensen et al. 2016, 2018; LaFree et al. 2018) have
found that non-ideological criminal behavior prior to radicalization often distinguishes violent from non-violent extremists. For example, in a multivariate study,
LaFree et al. (2018) discovered that U.S. extremists with criminal backgrounds
were significantly more likely to adopt violence after radicalizing than were extremists without crime histories. Similarly, using an expanded and updated sample,
Jensen et al. (2018) built on the LaFree et al. study and found that previous crime
was the strongest non-ideological predictor of post-radicalization violent extremism
when comparing more than 1800 violent and non-violent U.S extremists. Both the
LaFree et al. and Jensen et al. studies used large samples and controlled for a number of ideological and non-ideological factors, indicating that the link between pre-­
radicalization crime and post-radicalization violence is robust.
Crime Specialization and Diversity
Criminologists have long investigated whether criminals specialize in specific types
of criminal behavior or whether they are generalists who engage in diverse forms of
crime and delinquency. While not all research supports the vision of ordinary
While studies at the individual-level are rare, there are numerous studies at the group-level on the
terror-crime nexus (Hutchinson and O’malley 2007; Picarelli 2006, 2012; Sanderson 2004; Toros
and Mavelli 2013).
2
The Link Between Prior Criminal Record and Violent Political Extremism in the United… 125
c­ riminals as generalists (Kempf 1987; Lussier et al. 2005; Shover 1996), a growing
consensus in the research on criminal trajectories finds that while there is some
evidence of specialization, most criminal careers are marked by “versatile offending
patterns” (Blumstein et al. 1986, p. 455; Piquero et al. 2003; Wolfgang et al. 1972).
For example, in a study of a Philadelphia youth cohort, Piquero (2000) found that
there was no evidence for specialization in offending. Rather, some non-violent
offenders eventually engaged in acts of criminal violence in addition to increasing
their offending frequency as they grew older. However, more recent research suggests that offending over the criminal career may vary considerably (Deane et al.
2005; Shover 1996; Steffensmeier and Ulmer 2005; Sullivan et al. 2006). For example, it appears that both individual-level propensity to commit crime, as well as
changes in life circumstances (e.g., employment, marriage, drug and alcohol use),
impact patterns of offense specialization/versatility in the short term (McGloin
et al. 2007).
While research on criminal specialization may provide valuable insights into the
processes by which extremists evolve from engaging in more ordinary types of non-­
violent crime to acts that are intended to injure or kill, only a few terrorism studies
to date have touched on this issue. For example, in a study of 79 European jihadists
with criminal records, Basra and Neumann (2016) found that the majority had committed petty crimes prior to plotting domestic terrorist attacks or traveling to fight in
foreign conflicts. Similarly, Jensen et al. (2018) found that many individuals who
went on to commit acts of political violence in the United States had previous criminal records that were limited to non-violent crimes, such as theft and drug offenses.
Although these studies have begun to address the criminal life-course of extremists,
there is a near absence of information on the non-ideological crime patterns of those
who have committed terrorism. Thus, there are very few studies that provide criminal offending rates for political extremists or offer data on the types of crimes that
political extremists engage in prior to radicalizing. Moreover, there are few comparisons of the criminal pasts of non-violent and violent extremists, including
whether multiple offending or criminal diversity is more or less prevalent in their
backgrounds.
Age of Onset and Violent Crime
In the 1980s and 1990s, a great deal of criminological research concentrated on
understanding the impact of early offending on the nature, severity, and duration of
criminal activity over an individual’s life-course (Blumstein et al. 1986; Farrington
1992; Laub and Sampson 1993). Unsurprisingly, the results of this research showed
that the age of onset of criminal activity is a reliable predictor of a number of future
criminal outcomes. For instance, in Glueck and Glueck’s (1950) classic analysis of
500 boys growing up in Boston, the authors found that age of onset was significantly related to both criminal career length and frequency of offending. Further, in
Wolfgang et al.’s (1972) influential Philadelphia study, the authors found that early
126
M. A. Jensen et al.
age of onset was related to both the persistence and severity of offending. That is,
individuals who committed their first crime at an earlier age were significantly more
likely to commit serious offenses in the future. The relationship appears to be so
robust across studies and variables (Hingson et al. 2006; Kaplow and Widom 2007;
Robins and Przybeck 1985) that some scholars declare age of onset to be the best
predictor of the future course of one’s criminal career (Blumstein et al. 1985;
Gottfredson and Hirschi 1990).
While there is a growing body of terrorism research that looks at the recruitment
of children into extremist organizations (Beber and Blattman 2013; Rosen 2005),
we are unaware of any studies that examine the relationship between juvenile non-­
ideological crime and later involvement in violent extremism.
Research Questions
While the previous studies (Jensen et al. 2016, 2018; LaFree et al. 2018) that have
explored pre-radicalization criminality and post-radicalization violent extremism
offer promising insights into the proccesses of radicalization to violence, important
research gaps remain. In particular, past studies have not estimated the strength of
the association between past non-ideological crimes and acts of political violence,
relying instead on descriptive analyses and case studies. Nor has extant research
analyzed the particular types of crime or offender characteristics that may be linked
to a greater propensity for violent extremism. To fill these gaps, this chapter
addresses the following questions:
• How does the risk calculus for violent extremism change if an individual has a
history of non-ideological criminal behavior? Is an individual with a history of
non-ideological violent offending at a greater risk of engaging in violent extremism once they have radicalized?
• Are there significant cross-ideological differences in the criminal histories of
U.S. extremists?
• Which pre-radicalization non-ideological criminal behaviors are most prevalent
among extremists with criminal histories?
• Are individuals who commit non-extremist crimes as juveniles at an increased
risk for engaging in violent extremism once they have radicalized?
Data
Data for this study were drawn from the Profiles of Individual Radicalization in the
United States (PIRUS) dataset, which contains individual-level information on the
backgrounds, attributes, and radicalization processes of 1867 violent and non-­
violent U.S. political extremists. The PIRUS data include individuals who adhered
The Link Between Prior Criminal Record and Violent Political Extremism in the United… 127
to far-left, far-right, Islamist, or single-issue ideologies and radicalized in the United
States from 1948 to 2016.3 The PIRUS data were collected using publicly available
sources, with priority given to information from well-regarded and reliable sources,
such as professional news media outlets, unsealed court records, and unclassified
government reports. In addition, researchers used websites that had uncertain reliability but contained links to reliable documents, such as court transcripts.
START researchers compiled the PIRUS dataset over multiple waves of data collection between 2013 and 2017. During the data collection phase, trained researchers double-coded approximately half of the sample (described below) in order to
increase the overall quality of the data, as well as to assess the reliability of the
coding instrument. At the end of each coding wave, researchers performed
Krippendorf’s alpha procedure to test for inter-rater reliability across the double-­
coded cases (Krippendorf and Hayes 2007). Each wave scored above a 0.7, which
indicates acceptable reliability for social science applications. After conducting reliability tests, project leaders reviewed all instances of discrepant coding and made a
final coding decision that best represented information available in the open sources
and most closely aligned with definitions laid out in the PIRUS codebook.
To be included in the PIRUS dataset, individuals must meet three requirements:
(1) their radicalization began or occurred primarily in the United States; (2) they
espoused clearly identifiable ideological motives; and (3) their terrorism related
behavior was linked to the ideological motives they espoused.4 In addition to these
criteria, individuals must meet at least one of the following criteria: (4a) they were
arrested; (4b) indicted of a crime; (4c) killed as a result of their ideological activities; (4d) were a member of a designated terrorist organization; or (4e) were associated with an organization whose leader(s) or founder(s) has been indicted of
ideologically motivated violent offenses.
Many studies of political violence based on open source data have substantial
amounts of missing data on key variables (Chermak et al. 2012; Dugan and Distler
2016; LaFree et al. 2018; Safer-Lichtenstein et al. 2017). While our main dependent
variable—violent/non-violent—and several other variables (gender, ideology, and
exposure year) include no missing values, the remaining variables have a wide
range of missing data. At the most extreme, one variable—presence of a radical
family member—is missing data in more than 80% of the cases. Three additional
variables (stable employment history, education level, and presence of rival groups)
are missing in more than 60% of the cases. For the descriptive analysis we simply
exclude cases with missing data. For our multivariate analysis we estimate models
using regression-based multiple imputation (Rubin 2004). This process is described
in greater detail below.
3
For a detailed explanation of these ideological categories, see http://www.start.umd.edu/sites/
default/files/files/research/PIRUSCodebook.pdf
4
For additional details on the PIRUS dataset and data collection methodology, please refer to the
project’s
final
report,
available
at
https://www.start.umd.edu/pubs/START_NIJ_
EmpiricalAssessmentofDomesticRadicalizationFinalReport_Dec2016_0.pdf
128
M. A. Jensen et al.
Of the individuals in PIRUS, we classified 1103 (59.1%) as violent extremists,
defined as individuals who conspired to or committed acts intended to kill or injure,
while 764 (40.9%) were coded as non-violent extremists, defined as individuals
who limited their extremist acts to those which did not threaten others’ physical
wellbeing, including vandalism, illegal protest, material support, and the destruction of uninhabited property.
A subset of 397 (21.3%) individuals in PIRUS were coded as having pre-­
radicalization criminal histories, which we define as any criminal act (aside from
non-felony drug use or moving traffic violations) with a non-ideological motive that
was committed before an individual adopted an extremist belief system. Neither
arrest nor conviction were required for an individual to be coded as committing pre-­
radicalization criminal offenses, although in most cases the individual did have
some kind of interaction with the criminal justice system. Individuals were also
coded as having a criminal history if they self-reported engaging in criminal acts
prior to radicalizing, even if they were never arrested or charged for a crime. In
order to address the research questions described above, we added several additional variables to the PIRUS data for each individual in the subsample:
• Previous criminal activity: an ordinal variable that records the most extreme
crime that an individual committed prior to radicalizing: 0 = no previous criminal history, 1 = misdemeanor non-violent crime; 2 = felony non-violent crime;
3 = any violent crime.
• Criminal activity type: a categorical variable that captures the nature of criminal
offenses that an individual committed prior to radicalizing.5
• Criminal activity age: a categorical variable that captures the age at which an
individual was first known to engage in non-ideological crime: 1 = under 18 years
old, 2 = 18–24 years old, 3 = 25–34 years old, 4 = 35–44 years old, 5 = 45 years
old and older.
Descriptive Results
We begin with a descriptive analysis of crime rates for individuals in PIRUS, as well
as a comparison of the criminal histories of perpetrators across different ideological
milieus. This allows us to provide a detailed examination of the sub-ideological
affiliations of individuals in PIRUS with histories of pre-radicalization crime. We
then analyze the specific types of crime that were most common in the pre-­
radicalization histories of the individuals in PIRUS, paying particular attention to
the crimes that were most prevalent in the backgrounds of violent extremists.
Finally, we examine the nexus between age of criminal onset and the likelihood that
individuals engage in acts of political violence after they have adopted an extremist
If an individual committed more than three crime types, coders were instructed to select the three
most serious offenses and list the remaining offenses in an additional text box.
5
The Link Between Prior Criminal Record and Violent Political Extremism in the United… 129
belief system. In particular, we assess whether there is a significant link between
juvenile non-ideological delinquency and post-radicalization violent extremism.
Pre-radicalization Crime Rates of U.S. Extremists
As noted above, we found that 397 (21.3%) subjects engaged in at least one criminal
act prior to radicalizing. Of the remaining cases, 608 (32.5%) were confirmed to
have no prior criminal record, while 862 (46.2%) were coded as “unknown” for
criminal history due to incomplete source documents. Below we discuss in greater
detail the strategies we use to compensate for these missing values and how they
impact our results. Of the 397 individuals in PIRUS who were coded as having pre-­
radicalization criminal records, 200 (50.4%) engaged exclusively in non-violent
crimes, such as drug offenses, theft, and vandalism. The remaining individuals were
divided between those who engaged exclusively in violent criminal acts (n = 78;
39.6%) and those who committed a mix of violent and non-violent crimes (n = 119;
60.4%). Nearly half (48.9%) of the individuals in PIRUS with criminal histories
engaged in multiple crimes prior to engaging in ideologically motivated criminal
behavior.
In Fig. 1, we compare non-violent and violent political extremists who have no
criminal history, a non-violent criminal history, and a violent criminal history.
Approximately 44% of the violent extremists and 32% of the non-violent extremists
in PIRUS committed at least one criminal offense prior to radicalizing. Further, the
data reveal important differences in the criminal histories of violent and non-violent
extremists. For example, individuals who went on to engage in acts of violent
extremism after radicalizing were significantly more likely to have committed acts
67.81%
56.07%
No Criminal History
Non-Violent Criminal History
24.44%
20.58%
19.49%
11.61%
Non-Violent Extremists (n=764)
Violent Extremists (n=1103)
Fig. 1 Rates of pre-radicalization crime in the PIRUS database
Violent Criminal History
130
M. A. Jensen et al.
Table 1 Rates of criminal histories among individuals in PIRUS (by primary ideology)
No criminal history
History of misdemeanor non-violent crime
History of felony non-violent crime
History of violent crime
Far-right
45.76%
13.28%
14.12%
26.84%
Far-left
71.93%
8.77%
4.09%
15.20%
Islamist
65.96%
9.42%
7.60%
17.02%
Single-issue
70.20%
10.60%
5.96%
13.25%
of violent crime prior to radicalizing than their non-violent extremist counterparts.
Nearly 25% of the violent extremists in PIRUS had pre-radicalization criminal histories that included acts of violence, whereas violent crime was only present in the
criminal histories of 12% of non-violent extremists.
In Table 1, we compare the criminal histories of the individuals in PIRUS. The
results show considerable variation across our four primary ideological categories:
far-right, far-left, Islamist, and single-issue extremists.6 According to Table 1, far-­
right extremists in the PIRUS data display the highest rates of pre-radicalization
crime, including the highest rates of previous violent offenses. While only 54% of
far-right extremists in PIRUS had no criminal history, 72% of far-left extremists,
70% of single-issue extremists, and 66% of Islamists had no prior criminal activity.
Approximately 27% of far-right extremists in the data had criminal histories that
included acts of violent crime, such as homicide, armed assaults, and sexual violence. By comparison, 17% of Islamists, 15% of far-left extremists, and 13% of
single-issue extremists had criminal histories that included violent crime.
Interestingly, while individuals on the far-left commonly engaged only in acts of
non-violent political extremism after radicalizing (64.8%), those with criminal histories often (53.5%) committed acts of violent non-ideological crime prior to radicalizing. Nearly 35% of the Islamist extremists in PIRUS engaged in criminal acts
prior to radicalizing. Those with pre-radicalization criminal histories were evenly
split between those whose criminal acts were limited to non-violent offenses and
those whose crimes included acts of violence. Finally, 30% of individuals in PIRUS
who were classified as single-issue extremists had histories of criminal behavior
prior to radicalizing, including 13% who committed acts of violent crime.
In Table 2, we examine the extremists in PIRUS with criminal histories according to their sub-ideological affiliations. According to Table 2, the high rate of previous criminality on the far-right is primarily driven by individuals who harbored
white supremacist or Neo-Nazi views. Indeed, nearly 65% of far-right extremists in
PIRUS with previous criminal records eventually radicalized to adopt beliefs of
white racial supremacy. Extremists who expressed anti-government views, such as
Sovereign Citizens, also constitute a substantial portion (32.8%) of the far-right
cases in PIRUS with previous criminal histories.
Similar patterns of previous criminality are also found for the other sub-­
ideologies. For example, individuals affiliated with the Black nationalist/separatist
movements of the 1960s and 1970s make up the largest portion (50%) of extremists
6
For detailed descriptions of the ideological categories, see Jensen et al. (2016).
The Link Between Prior Criminal Record and Violent Political Extremism in the United… 131
Table 2 # of individuals in PIRUS with criminal histories (by sub-ideology)
Ideology/sub-ideology
Far-right
Militia/gun rights
White supremacist/KKK/Neo-Nazi
Xenophobic/anti-immigrant
Anti-government/Sovereign Citizens movement
Christian identity
Far-left
Animal rights/environmentalist
New Left (1960’s student movements/anti-Vietnam
War)
Black nationalist/black separatist
Anti-capitalist/communist/anti-imperialist
Anarchist
Islamist
Al-Qaeda affiliated/inspired
Islamic State affiliated/inspired
Unaffiliated Salafist/other
Single-issue
Puerto Rican independence/Puerto Rican nationalist
Cult/idiosyncratic
Anti-abortion
Anti-gay
Other
# of individuals with criminal
history
192
22
124
13
63
10
48
11
7
24
6
7
112
34
41
37
45
1
2
21
8
18
Note: Subjects can be coded for more than one ideological subcategory
on the far-left with pre-radicalization criminal histories. Islamist extremists in
PIRUS with pre-radicalization criminal backgrounds appear to be evenly divided
between those who aligned with or were inspired by Al-Qaeda and its associated
movements (AQAM) and those who aligned with or were inspired by the Islamic
State of Iraq and Syria (ISIS; 30.3% and 36.6% respectively). Previous criminality
among single-issue extremists in PIRUS is most prevalent in the backgrounds of
anti-abortion extremists.
Pre-radicalization Criminal Activity of U.S. Extremists
Figure 2 shows the range of criminal activities that the perpetrators in PIRUS engaged
in prior to radicalizing. These behaviors included a number of violent offenses, such
as homicide, assault, and sexual violence, and a long list of non-violent crimes,
including drug offenses, theft, vandalism, and fraud.
132
M. A. Jensen et al.
Aggravated Assault
Drug Related
Simple Assault
Illegal Use/Possession of Firearm
Robbery
Larceny-Theft
Burglary
Fraud
Driving Under the Influence (DUI)
Homicide
Vandalism
Motor Vehicle Theft
Arson
Forcible Rape
Forgery
Prostitution
Parole Violation
Embezzlement
22.92%
18.14%
15.11%
14.61%
14.11%
12.34%
9.82%
7.81%
7.05%
5.29%
3.53%
Violent Crime
Non-Violent Crime
3.27%
3.02%
2.52%
1.26%
1.01%
0.76%
0.50%
0.00%
5.00%
10.00% 15.00% 20.00% 25.00%
Fig. 2 Involvement in pre-radicalization criminal behaviors among U.S. extremists. (Note: Arson
can be classified as either a violent or non-violent crime depending on whether the intention of the
act was to kill or injure)
Aggravated assault was the most common crime type for those in the PIRUS data
who engaged in criminal acts prior to radicalizing. Nearly 23% of extremists with
criminal histories in PIRUS committed at least one aggravated assault before radicalizing. Aggravated assaults were more common in the criminal histories of violent
extremists, with nearly 27% engaging in this crime versus 14% for non-violent
extremists. The rates of aggravated assault were similar across the ideological
groups in PIRUS. Just over 22% of Islamists, 25% of far-right extremists, and 27%
of far-left extremists with criminal histories committed at least one aggravated
assault prior to radicalizing.
Of the extremists in the PIRUS data who had a history of violent crime prior to
radicalizing, 12.7% committed acts of domestic violence (although this may be a
conservative estimate given that instances of domestic violence may not be reported
or may appear in sentencing records only as assault and/or battery charges). Other
violent crimes found in the backgrounds of the individuals in PIRUS include simple
assault (15.1%), robbery (14.1%), homicide (5.2%), and sexual assault (2.5%).
Of the violent extremists in PIRUS with criminal histories, 44.3% limited their
criminal activities to non-violent offenses. The most common non-violent crime in
the backgrounds of the individuals in PIRUS was the use, possession, or distribution
of illicit drugs. In fact, drug-related charges were the second most common pre-­
radicalization crime type for extremists in the PIRUS data with criminal histories.
More than 18% of the extremists in the data with criminal histories engaged in at
least one illegal act involving illicit drugs prior to radicalizing. Drug offenses were
relatively more common among violent extremists in the data: over 20% of violent
extremists with criminal histories had drug offenses on their criminal records compared to 13% for non-violent extremists.
The Link Between Prior Criminal Record and Violent Political Extremism in the United… 133
The individuals in PIRUS engaged in a number of other non-violent criminal
acts, including illegal firearms possession (20.4%), fraud (14.7%), theft (13.9%),
and driving under the influence of alcohol (9%). Larceny-theft was the single most
common pre-radicalization crime type for non-violent extremists in PIRUS. Nearly
15% of non-violent extremists in the database committed at least one known theft
prior to radicalizing. This outpaced other common crime types for non-violent
extremists, including aggravated assault (14%), drug offenses (13%), and simple
assault (12%).
The pre-radicalization criminal behaviors of a small percentage (14.1%) of the
extremists in PIRUS were tied to their involvement in street gangs or criminal organizations.7 Of the individuals with past ties to street gangs, approximately 80% went
on to engage in acts of violent extremism after radicalizing.
Taken together, the data on pre-radicalization crime patterns and offenses in
PIRUS suggest that political extremists rarely specialize in specific crime types. The
individuals that we examined who committed criminal offenses prior to radicalizing
engaged in a wide range of criminal acts, from less serious forms of delinquency,
such as theft and drug possession, to more serious forms of violent crime, including
assaults and homicides. Moreover, many individuals in the PIRUS data gradually
escalated in their criminal activities from non-violent offenses prior to radicalizing
to violent acts after adopting extremist beliefs. By contrast, some individuals, especially those affiliated with far-left groups, limited their extremist activities to non-­
violent crimes, even though many committed violent offenses prior to radicalizing.
Age of Crime Onset
In Fig. 3, we list the average age of earliest criminal activity for individuals with
criminal records for both violent and non-violent political extremists. The PIRUS
data show that, on average, U.S. extremists are well into adulthood when they first
engage in ideologically motivated acts. For example, violent extremists in PIRUS
were on average 33 years old when they first participated in illegal acts on behalf of
their extremist beliefs. However, the same individuals in PIRUS were often much
younger when they first engaged in non-ideological crime. In fact, 62.1% of the
extremists in PIRUS with criminal histories first engaged in non-ideological crimes
when they were 24 years old or younger. Most commonly, the extremists in PIRUS
first committed non-ideological crimes when they were between 18 and 24 years
old. Specifically, 44.6% of violent extremists and 43.9% of non-violent extremists
in PIRUS with criminal histories first engaged in criminal acts when they were
18–24 years old.
See Pyrooz et al. (2017) for a detailed study of the connections between gang involvement and
extremism in the PIRUS dataset.
7
134
M. A. Jensen et al.
44.64% 43.87%
22.92%
21.74%
18.75%
16.96%
9.82%
8.30%
9.82%
3.16%
Under 18 years old
18-24 years old
25-34 years old
Non-Violent Extremists
35-44 years old
45+ years old
Violent Extremists
Fig. 3 Average age of earliest non-ideological criminal activity
While less common, the rates of juvenile crime among the individuals in the
PIRUS data nonetheless reveal important differences in the backgrounds of violent
and non-violent extremists. Approximately 17% of the extremists in PIRUS with
pre-radicalization criminal histories first engaged in crime when they were under
the age of 18 years old. However, those who went on to commit acts of violent
extremism show significantly higher rates of juvenile offending. Nearly 22% of
violent extremists with criminal histories began engaging in criminal acts when they
were under the age of 18. In comparison, less than 10% of non-violent extremists
with criminal histories engaged in crime when they were juveniles.
Multivariate Analysis
To better assess the relationship between pre-radicalization crime and post-­
radicalization violence, we next performed a multivariate logit regression analysis
that estimates how the likelihood of violent extremism changes if individuals engage
in crime prior to radicalizing. Before moving on to the results of this analysis, we
first provide brief descriptions of the dependent, independent, and control variables
that are used in the model. We have already described the main dependent variable,
violent, coded “1” for individuals whose earliest public exposure involved an act of
violence or the intention to commit an act of violence, and “0” for individuals whose
earliest public exposure involved only non-violent acts.
Our main independent variable for this part of the analysis is prior criminal activity, which is coded “1” for participation in violent crime, “0.5” for participation in
non-violent crime, and “0” for no criminal participation prior to involvement in
extremist activities.
The Link Between Prior Criminal Record and Violent Political Extremism in the United… 135
Control Variables
For our multivariate analysis, we include a number of theoretically relevant control variables drawn from prior research. Stable employment is coded “1” for
individuals who worked regularly prior to engaging in acts of extremism and “0”
for individuals who were unemployed, alternated between periods of employment
and unemployment, or habitually changed careers in the years leading up to their
involvement in extremism. Education is an ordinal measure that captures the highest level of education attained by individuals at the time they engaged in illegal
acts. Individuals were coded “3” if they received a college degree or higher, “2” if
they attended college but did not graduate, “1” if they finished high school, or “0”
if they did not graduate from high school. Marital status is coded “1” for individuals with a legally-recognized spouse or domestic partner and “0” for divorced or
widowed individuals and those who never married. Individuals who were once
active in the military but were discharged before they engaged in extremism were
coded “1” for past military experience and “0” otherwise. Those on active military
duty at the time of their public exposure were coded “1” for active in military and
“0” otherwise.
We coded individuals who were part of close-knit, insular cliques with others
who shared similar extremist views as “1” for radical peers and “0” otherwise. We
coded individuals who had immediate or extended family members who participated in illegal non-violent or violent ideologically-motivated behavior as “1” for
radical family and “0” otherwise.
We coded mental illness as “1” if there is evidence that an individual suffered
from mental illness of any kind, including schizophrenia, bipolar disorder, post-­
traumatic stress disorder, or major depression and “0” otherwise. We coded mental
illness as positive based on either clinical diagnosis or self-reports and testimony by
family or friends.
To control for the possibility that individuals are more likely to engage in violence when they are members of competing groups, we coded rival groups as “1”
for individuals who were members of an extremist group that suffered from internal splintering or competed with like-minded organizations for status, prestige, or
resources and “0” otherwise. Evidence of internal competition included fractionalization within the group, acts or threats of violence between group members,
and leadership turnover due to disagreements about group behavior or goals. We
also coded this variable positively if there was evidence of violence between
rival groups.
To control for violence-justifying beliefs that may be present across the political
spectrum we included several ideology measures. These measures are binary variables that categorize individuals along far right, far left, radical Islamist, or single-­
issue ideological milieus (the categories are mutually exclusive). Gender is coded as
a dichotomy (“1” = men; “0” = women) and age is a continuous measure recorded
at the time at which the individual’s extremist activities first became public knowledge (e.g., they were arrested, their plot materialized).
136
M. A. Jensen et al.
Missing Data
Several methods exist for mitigating the problem of missing data in large datasets,
including simply excluding cases that have missing values from the analysis (i.e.,
list-wise deletion), replacing missing data with a universal baseline value (i.e.,
fixed-value imputation, or cold-deck imputation), or performing complex statistical
models for estimating missing values based off key extant data (see Safer-­
Lichtenstein et al. 2017).
Based on our previous work (Jensen et al. 2016, 2018; LaFree et al. 2018), we
relied on a regression-based multiple imputation technique to predict missing values. Regression-based multiple imputation operates on the assumption that data are
missing conditionally at random (Rubin 2004). Thus, while the missing information
for a given variable may not be randomly dispersed among all observations in a
study, it is reasonable to assume a certain degree of randomness to missing values
when accounting for other potentially confounding variables. We determined that
this is a safe supposition given the wealth of information contained in the PIRUS
dataset. Based off precedent in other terrorism research (Gruenewald and Pridemore
2012) and established practices in dealing with missing data in social science
research (Rubin 2004), we produced 50 imputed datasets using regression techniques and report the pooled values below.
Multivariate Results
Table 3 shows the results of the multivariate logistical analysis. A number of variables are significantly associated with violent or potentially violent extremist outcomes (as opposed to non-violent extremism). Perhaps most importantly, the results
indicate that pre-radicalization criminal activity, violent or non-violent, is the single
strongest predictor of post-radicalization violent extremism when controlling for
ideology.8 In fact, individuals who engaged in non-violent or violent crime prior to
radicalizing were 1.85 times more likely to engage in acts of violent extremism after
radicalizing than were extremists without criminal histories. The effect of pre-­
radicalization criminal behavior on post-radicalization violence are the strongest for
extremists whose criminal histories include acts of violent crime (e.g., homicide,
Given that the goal of this study is to isolate the effect of pre-radicalization crime on the propensity for post-radicalization violence, ideology is treated here as a control variable. However, it is
worth noting that ideological affiliation is the single strongest predictor of violent outcomes in
PIRUS. Individuals associated with far-right and Islamist groups are substantially more likely to
participate in acts of violent extremism than are individuals on the extremist far-left. This result is
not surprising. Groups on the extremist far-right and those affiliated with Salafi jihadism actively
encourage their followers to conduct acts of violence on behalf of their respective movements,
while many far-left extremist groups promote the use of non-violent political resistance to achieve
their goals.
8
The Link Between Prior Criminal Record and Violent Political Extremism in the United… 137
Table 3 Multivariate logistic regression: criminal history (all) and participation in extremist
violence
Variable
Prior criminal activity
Stable employment history
Education
Married
Past military experience
Active in military
Radical peers membership
Radical family
Mental illness
Rival groups
Gender
Age
Age (squared)
Islamist ideology
Far right ideology
Far left ideology
Exposure 1950s
Exposure 1960s
Exposure 1970s
Exposure 1980s
Exposure 1990s
Exposure 2000s
Β
.618∗∗∗
−.450∗
−.006
−.302∗
.280
.212
.599∗∗∗
−.505∗
.521∗∗
−.024
.400∗
−.060∗
.001∗
1.064∗∗∗
.533∗∗
−1.023∗∗∗
.283
.171
.554∗
.353
−.300
−.681∗∗∗
SE
.187
.197
.106
.139
.206
.307
.150
.205
.203
.165
.198
.024
.000
.202
.169
.198
.576
.263
.224
.202
.182
.154
Odds ratio
1.855
.637
.994
.740
1.324
1.236
1.821
.604
1.684
.976
1.492
.942
1.001
2.897
1.704
.359
1.327
1.186
1.741
1.423
.741
.506
95% C.I. for odds ratio
Lower
Upper
1.284
2.681
0.424
0.958
0.796
1.241
0.562
0.974
0.883
1.983
0.677
2.255
1.349
2.458
0.401
0.909
1.131
2.507
0.699
1.363
1.012
2.200
0.899
0.986
1.000
1.001
1.950
4.302
1.224
2.372
0.244
0.529
0.429
4.104
0.709
1.985
1.122
2.702
0.957
2.115
0.519
1.058
0.375
0.684
DV = earliest public exposure in extremist activity involved intentional violence against a person
or persons (1), or earliest public exposure in extremist activity did not involve intentional violence
against a person or persons (0)
N = 1867
∗p ≤ .05; ∗∗p ≤ .01; ∗∗∗p ≤ .001 (two-tailed)
assault, forcible rape). Individuals with violent criminal histories were 2.44 times
more likely to engage in acts of violent extremism once they had radicalized.
We note in passing a number of control variables that were also significantly
linked to greater propensities for violence among the subjects in PIRUS.9 In addition to criminal history, individuals with a documented or suspected mental illness,
those who were members of extremist cliques, and men were all significantly more
likely to engage in acts of violent extremism. Conversely, individuals with stable
employment histories were nearly 35% less likely to engage in acts of violent
While not the focus of our analysis, it is worth noting that the presence of radical family members
appears to have a negative effect on the propensity of individuals to engage in violent extremism.
This counterintuitive finding appears to be driven by the large percentage (60%) of individuals on
the far-left who had radical family members and significant others at the time of their involvement
in extremist activities.
9
138
M. A. Jensen et al.
Table 4 Multivariate logistic regression: criminal history (juvenile) and participation in extremist
violence
Variable
Juvenile crime
Stable employment history
Education
Married
Past military experience
Active in military
Radical peers membership
Radical family
Mental illness
Rival groups
Gender
Age
Age (squared)
Islamist ideology
Far right ideology
Far left ideology
Exposure 1950s
Exposure 1960s
Exposure 1970s
Exposure 1980s
Exposure 1990s
Exposure 2000s
Β
.889∗∗
−.664∗
−.102
−.389∗
.292
.317
.471∗∗∗
−.444∗
.595∗∗
−.038
.410∗
−.051∗
.001∗
1.076∗∗∗
.526∗∗
−.959∗∗∗
.304
.037
.384
.265
−.444∗
−.764∗∗∗
SE
.356
.171
.089
.171
.208
.310
.146
.192
.202
.183
.195
.023
.000
.204
.174
.207
.578
.262
.222
.204
.183
.154
Odds ratio
2.433
.515
.903
.678
1.339
1.373
1.602
.642
1.814
.963
1.506
.951
1.001
2.933
1.692
.383
1.355
1.038
1.469
1.303
.642
.466
95% C.I. for odds ratio
Lower
Upper
1.205
4.911
0.353
0.749
0.757
1.076
0.483
0.949
0.891
2.010
0.747
2.522
1.203
2.134
0.438
0.938
1.219
2.696
0.670
1.383
1.028
2.206
0.907
0.995
1.000
1.001
1.965
4.377
1.202
2.381
0.255
0.575
0.436
4.207
0.621
1.734
0.951
2.267
0.874
1.942
0.448
0.918
0.344
0.629
DV = earliest public exposure in extremist activity involved intentional violence against a person
or persons (1), or earliest public exposure in extremist activity did not involve intentional violence
against a person or persons (0)
N = 397
∗p ≤ .05; ∗∗p ≤ .01; ∗∗∗p ≤ .001 (two-tailed)
extremism than individuals who were long-term unemployed. Extremists who were
older and married were also less likely to engage in acts of violent extremism. In
general, these results track very closely with our earlier work on the PIRUS data
(Jensen et al. 2018; LaFree et al. 2018).
We next considered whether juvenile offending is also predictive of post-­
radicalization violence. The results are shown in Table 4. In general, the results
indicate that juvenile offending is one of the strongest non-ideological predictors of
post-radicalization violence when controlling for other important correlates of
extremism. According to Table 4, when we reproduce the earlier multivariate analysis using juvenile crime as the measure of previous criminality, we find a significant
increase in the probability that an individual will engage in acts of violent e­ xtremism
after radicalizing. Indeed, when juvenile offending is used as the measure of previ-
The Link Between Prior Criminal Record and Violent Political Extremism in the United… 139
ous criminality in the multivariate model, the probability that an individual with a
criminal history prior to radicalizing will go on to engage in acts of violent extremism increases from 1.85 times more likely to 2.43 times more likely. As Table 4
shows, while several variables remain significantly related to a higher probability of
violent extremism, juvenile offending is the strongest non-ideological predictor of
post-radicalization violence among individuals in PIRUS. Importantly, the measure
of juvenile crime used in the model includes acts of violent and non-violent crime,
indicating that the link between early crime and violent extremism is not simply
limited to a continuation of violent deviant behavior in the individuals’ adult years.
Limitations
Before moving on to a discussion of the policy implications of these results, we
acknowledge several limitations in the present study. First, the variables included in
our study have varying rates of missing data, including some variables (e.g., employment history) that are missing values for more than 60% of the cases in
PIRUS. Missing data are a common feature of open source studies of political
extremism (Dugan and Distler 2016; Freilich et al. 2014; Safer-Lichtenstein et al.
2017). Important information may simply be unavailable in court documents and
open source media. In addition, it is likely that the cases with the least missing data
are those most prominently covered by news agencies. In the study of political
extremism, like the study of crime more generally, we face methodological trade-­
offs. While the limits of open source data in this area are clear, more traditional
methods like surveys and case studies also have corresponding weaknesses.
Nevertheless, considering the non-trivial amount of missing data in PIRUS, we caution that our results are best viewed as preliminary.
Second, while we argued above that contrasting violent and non-violent political
extremists has advantages, we are not able to distinguish those who radicalize to
commit illegal acts of political extremism from the general population. While such
a study would undoubtedly make significant contributions to political violence
research, it is simply not possible given current data limitations.
Third, while many theories of radicalization to violence (Borum 2011; Horgan
2008; McCauley and Moskalenko 2011; Neumann 2013) conceptualize it as the
culmination of multiple, dynamic pathways, our data do not allow us to measure
with precision the temporal ordering of the variables we include. Furthermore, the
data that we rely on do not allow us to address important debates in criminology
about the mechanisms that link individual causal factors to specific outcomes. While
not the focus of this study, we caution that our results do not reveal if individual
correlates are linked to violent extremism through selection effects, conjunctural
dynamics, or another type of causal process.
And finally, given that our sample is limited to U.S.-based extremists, we cannot
generalize our results to other countries or regions. While the theoretical ­perspectives
that we draw from are designed to be broad in explanatory scope, it could be that
140
M. A. Jensen et al.
variations in cultural, legal, and social norms will influence how well individual
criminological correlates explain violent extremism outside of the United States.
Future research should explore how well dominant criminological theories explain
violent extremism in other geographical contexts, especially those outside the West,
in order to better understand the benefits and limitations of applying criminological
theory to the study of extremism.
Discussion
Policy makers seeking to reduce the threat of violent extremism would benefit from
building on effective criminological policies used to make informed law enforcement decisions and to shape prevention and rehabilitation programs. To this end, our
results suggest several specific options. First, the results underscore the importance
of leveraging knowledge about an individual’s criminal record in prioritizing counterterrorism investigations. Specifically, the violent extremists in our sample were
more likely than the non-violent extremists to have a history of criminal behavior,
including both violent and non-violent offenses. It is important to note that while
our results cannot definitively dismiss the possibility that extremists specialize in
violent crimes throughout their criminal careers, preliminary evidence suggests that
extremists often progress from minor to more serious offending. If this finding is
confirmed in subsequent research, it provides useful information for identifying
individuals who are at risk of engaging in violent political extremism.
Second, our results are in line with a long criminological research tradition that
underscores the stigmatizing effects of prior criminal activity. Prior research shows
the cumulative disadvantage of a criminal history, including effects on perceived
worth (Evans et al. 2018; Gendreau et al. 1979; Luoma et al. 2007), ability to vote
(Burch 2011; Manza and Uggen 2008; Uggen and Manza 2002), others’ perceptions
(Braithwaite 1989; Matsueda 1992), job and housing prospects (Agan and Starr
2016; D’Alessio et al. 2015; Stoll and Bushway 2008), and subsequent physical and
mental health impediments (Freudenberg 2002; Schnittker and John 2007). Further,
there is evidence that stigmatization may be especially pertinent for terrorist perpetrators who are attempting to regain standing following their disengagement from
extremist activities (Jasko et al. 2016; Kruglanski et al. 2014). The results generally
support the utility of policies aimed at reducing the stigmatization of criminal histories and promoting reintegration into society. Moreover, this suggests that an added
benefit of effective rehabilitation programs for non-extremist offenders may be their
role in preventing future violent extremism.
Third, our study identified differences in the prevalence of prior criminal activity
when comparing individuals who ascribe to different extremist ideologies.
Specifically, individuals ascribing to a far-right ideology were more likely to have a
history of criminal activity, including violent crime. By contrast, far-left extremists
were far less likely to have a pre-radicalization criminal history. These findings
­suggest that ideology should be explicitly considered in studying the varied path-
The Link Between Prior Criminal Record and Violent Political Extremism in the United… 141
ways to radical extremism: there appear to be different risk factors for violent
extremism depending on an individual’s ideological milieu. Thus, our results suggest that criminal history is more salient for the far-right when compared to the
far-left or Islamist cases, a finding supported by related research on characteristics
of extremist perpetrators (Chermak and Gruenewald 2015; Gill et al. 2014).
Finally, our results show that violent extremists with a criminal history were
significantly more likely to have started their non-ideological offending at a younger
age than non-violent extremists. In other words, age of onset matters for entry into
violent extremism, as it does for important criminological outcomes, such as length
and severity of criminal careers. While we cannot identify specific causal mechanisms, our results show that early entry into crime is a risk factor for violent extremism. Intelligence practitioners would clearly do well to liaise with local law
enforcement and utilize existing criminal record databases to make informed risk
assessments. Perhaps more importantly, programs aimed at preventing violent
extremism in the United States stand to gain by partnering with existing criminal
justice programs that are in place to steer at-risk youth away from engaging in juvenile crime.
Conclusion
In general, criminologists have played a small but growing role in explaining the
etiology of violent political extremism. Our results suggest that criminological perspectives have a lot to offer to the study of political extremism and that they may
potentially provide answers to questions that have long puzzled terrorism scholars.
Indeed, political violence research has struggled to identify the risk factors that
explain why some individuals are willing to commit acts of violence on behalf of
their extremist belief system while others are not. The present study shows that
criminal record, which plays an especially important role in life course (Sampson
and Laub 1997), subcultural (Singer 1986) and labeling/conflict perspectives (Pager
2003), may be one important risk indicator that separates violent and non-violent
extremists. While the link between pre-radicalization criminality and post-­
radicalization violence appears to be robust, research gaps remain. We conclude
with a short discussion of the opportunities for future research.
First, while contrasting violent and non-violent political extremists among a
sample of ideologically committed individuals provides a useful comparison, future
studies should attempt to distinguish those who radicalize to commit illegal acts of
political extremism from the general population. Such a study would allow analysts
to make inferences about the specific mechanisms that link criminal history to radicalization for some individuals but not others. Second, while many theories of radicalization (Borum 2011; Horgan 2008; McCauley and Moskalenko 2011; Neumann
2013) conceptualize it as the culmination of multiple, dynamic pathways, the
empirical study of extremism more often treats the phenomenon as linear and ­simple
(for an exception, see Jensen, Atwell Seate, and James 2018). While we can estab-
142
M. A. Jensen et al.
lish that criminal activity often comes before radicalization, future studies would be
well served to examine the complex longitudinal nuances that are often present in
the radicalization pathways of extremists. This would include not only an assessment of the temporal proximity of previous crime to radicalization, but also how
previous criminal activity interacts with other criminological correlates to produce
unique pathways to extremist outcomes. Alternative types of data, such as panel or
time series, would be ideal to measure these types of temporal processes. And
finally, given that our sample is limited to U.S.-based extremists, future terrorism
studies could further estimate the robustness of the relationship between criminal
history and violent extremism by examining it in other countries or regions of
the world.
References
Agan, A. Y., & Starr, S. B. (2016). Ban the box, criminal records, and statistical discrimination: A
field experiment (SSRN Scholarly Paper No. ID 2795795). Rochester: Social Science Research
Network. Retrieved from https://papers.ssrn.com/abstract=2795795
Basra, R., & Neumann, P. R. (2016). Criminal pasts, terrorist futures: European jihadists and the
new crime-terror nexus. Perspectives on Terrorism, 10(6). Retrieved from http://www.terrorismanalysts.com/pt/index.php/pot/article/view/554.
Beber, B., & Blattman, C. (2013). The logic of child soldiering and coercion. International
Organization, 67(1), 65–104. https://doi.org/10.1017/S0020818312000409.
Blumstein, A., Farrington, D. P., & Moitra, S. (1985). Delinquency careers: Innocents, desisters,
and persisters. Crime and Justice, 6, 187–219.
Blumstein, A., Cohen, J., Roth, J. A., & Visher, C. A. (Eds.). (1986). Criminal careers and career
criminals (Vol. 2, 1st ed.). Washington, D.C: National Academies Press.
Blumstein, A., Cohen, J., & Farrington, D. P. (1988). Criminal career research: Its value for criminology. Criminology, 26(1), 1–35. https://doi.org/10.1111/j.1745-9125.1988.tb00829.x.
Borum, R. (2011). Radicalization into violent extremism I: A review of social science theories.
Journal of Strategic Security, 4(4), 7–36.
Braithwaite, J. (1989). Crime, shame and reintegration. New York: Cambridge University Press.
Brame, R., Bushway, S. D., Paternoster, R., & Turner, M. G. (2014). Demographic patterns of
cumulative arrest prevalence by ages 18 and 23. Crime and Delinquency, 60(3), 471–486.
https://doi.org/10.1177/0011128713514801.
Burch, T. (2011). Turnout and party registration among criminal offenders in the 2008 general election. Law & Society Review, 45(3), 699–730. https://doi.org/10.1111/j.1540-5893.2011.00448.x.
Chermak, S., & Gruenewald, J. A. (2015). Laying a foundation for the criminological examination
of right-wing, left-wing, and Al Qaeda-inspired extremism in the United States. Terrorism and
Political Violence, 27(1), 133–159. https://doi.org/10.1080/09546553.2014.975646.
Chermak, S. M., Freilich, J. D., Parkin, W. S., & Lynch, J. P. (2012). American terrorism and
extremist crime data sources and selectivity bias. Journal of Quantitative Criminology, 28(1),
191–218. https://doi.org/10.1007/s10940-011-9156-4.
Clark, D. (2017, August 18). How many U.S. adults have a criminal record? Retrieved January
7, 2019, from http://www.politifact.com/new-york/statements/2017/aug/18/andrew-cuomo/
yes-one-three-us-adults-have-criminal-record/
Clarke, R. V. G., & Newman, G. R. (2006). Outsmarting the terrorists. Westport: Praeger Security
International.
The Link Between Prior Criminal Record and Violent Political Extremism in the United… 143
Cottee, S. (2016, January 25). Reborn into terrorism. The Atlantic. Retrieved from http://www.
theatlantic.com/international/archive/2016/01/isis-criminals-converts/426822/
D’Alessio, S. J., Stolzenberg, L., & Flexon, J. L. (2015). The effect of Hawaii’s ban the box
law on repeat offending. American Journal of Criminal Justice, 40(2), 336–352. https://doi.
org/10.1007/s12103-014-9251-9.
Deane, G., Armstrong, D. P., & Felson, R. B. (2005). An examination of offense specialization using marginal logit models∗. Criminology, 43(4), 955–988. https://doi.
org/10.1111/j.1745-9125.2005.00030.x.
DeLisi, M., Neppl, T. K., Lohman, B. J., Vaughn, M. G., & Shook, J. J. (2013). Early starters:
Which type of criminal onset matters most for delinquent careers? Journal of Criminal Justice,
41(1), 12–17. https://doi.org/10.1016/j.jcrimjus.2012.10.002.
Dugan, L., & Distler, M. (2016). Measuring terrorism. In The handbook of the criminology of terrorism (pp. 189–205). New York: Wiley-Blackwell.
Evans, D. N., Pelletier, E., & Szkola, J. (2018). Education in prison and the self-stigma:
Empowerment continuum. Crime & Delinquency, 64(2), 255–280. https://doi.
org/10.1177/0011128717714973.
Farrington, D. P. (1992). Criminal career research in the United Kingdom. The British Journal of
Criminology, 32(4), 521–536. https://doi.org/10.1093/oxfordjournals.bjc.a048255.
Forst, B., Greene, J. R., & Lynch, J. P. (2011). Criminologists on terrorism and homeland security.
New York: Cambridge University Press.
Freudenberg, N. (2002). Adverse effects of US jail and prison policies on the health and well-­
being of women of color. American Journal of Public Health, 92(12), 1895–1899. https://doi.
org/10.2105/AJPH.92.12.1895.
Freilich, J., Chermak, S., Belli, R., Gruenwald, J., & Parkin, W. (2014). Introducing the extremist
crime database (ECDB). Terrorism and Political Violence, 26(2), 372–384.
Gendreau, P., Grant, B. A., & Leipciger, M. (1979). Self-esteem, incarceration and recidivism.
Criminal Justice and Behavior, 6(1), 67–75. https://doi.org/10.1177/009385487900600110.
Gendreau, P., Little, T., & Goggin, C. (1996). A meta-analysis of the predictors of adult offender recidivism: What works. Criminology, 34(4), 575–608. https://doi.org/10.1111/j.1745-9125.1996.
tb01220.x.
Gill, P., Horgan, J., & Deckert, P. (2014). Bombing alone: Tracing the motivations and antecedent
behaviors of lone-actor terrorists. Journal of Forensic Sciences, 59(2), 425–435. https://doi.
org/10.1111/1556-4029.12312.
Glueck, S., & Glueck, E. (1950). Unraveling Juvenile delinquency. Oxford, England:
Commonwealth Fund.
Gottfredson, M., & Hirschi, T. (1990). A general theory of crime (1st ed.). Stanford: Stanford
University Press.
Gruenewald, J., & Pridemore, W. A. (2012). A comparison of ideologically-motivated homicides
from the new extremist crime database and homicides from the supplementary homicide
reports using multiple imputation by chained equations to handle missing values. Journal of
Quantitative Criminology, 28, 141–162. https://doi.org/10.1007/s10940-011-9155-5.
Gruenewald, J., Chermak, S., & Freilich, J. D. (2013). Far-right lone wolf homicides in the United
States. Studies in Conflict & Terrorism, 36(12), 1005–1024. https://doi.org/10.1080/10576
10X.2013.842123.
Hingson, R. W., Heeren, T., & Winter, M. R. (2006). Age at drinking onset and alcohol dependence: Age at onset, duration, and severity. Archives of Pediatrics & Adolescent Medicine,
160(7), 739–746. https://doi.org/10.1001/archpedi.160.7.739.
Hoffman, P. B., & Beck, J. L. (1974). Parole decision-making: A salient factor score. Journal of
Criminal Justice, 2(3), 195–206. https://doi.org/10.1016/0047-2352(74)90031-2.
Hoffman, P. B., Stone-Meierhoefer, B., & Beck, J. L. (1978). Salient factor score and releasee
behavior: Three validation samples. Law and Human Behavior, 2(1), 47–62. https://doi.
org/10.1007/BF01047502.
144
M. A. Jensen et al.
Horgan, J. (2008). From profiles to pathways and roots to routes: Perspectives from psychology
on radicalization into terrorism. The Annals of the American Academy of Political and Social
Science, 618(1), 80–94. https://doi.org/10.1177/0002716208317539.
Hunter, R., & Heinke, D. (2011). Perspective: Radicalization of Islamist terrorists in the
Western World. Retrieved January 7, 2019, from https://leb.fbi.gov/articles/perspective/
perspective-radicalization-of-islamist-terrorists-in-the-western-world
Hutchinson, S., & O’malley, P. (2007). A crime–terror nexus? Thinking on some of the links
between terrorism and criminality. Studies in Conflict & Terrorism, 30(12), 1095–1107. https://
doi.org/10.1080/10576100701670870.
Jasko, K., LaFree, G., & Kruglanski, A. (2016). Quest for significance and violent extremism: The
case of domestic radicalization. Political Psychology., 38, 815–831. https://doi.org/10.1111/
pops.12376.
Jensen, M., James, P., LaFree, G., Atwell-Seate, A., Pisoiu, D., & Stevenson, J. (2016). Empirical
Assessment of Domestic Radicalization (EADR) (Final Report No. National Institute of Justice,
Award Number 2012-ZA-BX-0005). College Park: National Consortium for the Study of
Terrorism and Responses to Terrorism (START). Retrieved from https://www.ncjrs.gov/pdffiles1/nij/grants/250481.pdf
Jensen, M., James, P., & Safer-Lichtenstein, A. (2018). Pre-radicalization criminal activity of
United States extremists (Research Brief). College Park: The National Consortium for the
study of terrorism and responses to Terrorism. Retrieved from http://www.start.umd.edu/pubs/
START_PIRUS_PreRadCriminalActivityofUSExtremists_Jan2018.pdf
Jensen, M. A., Atwell Seate, A., & James, P. A. (2018). Radicalization to violence: A pathway
approach to studying extremism. Terrorism and Political Violence, 1–24.
Kaplow, J. B., & Widom, C. S. (2007). Age of onset of child maltreatment predicts long-term
mental health outcomes. Journal of Abnormal Psychology, 116(1), 176–187. https://doi.
org/10.1037/0021-843X.116.1.176.
Kempf, K. L. (1987). Specialization and the criminal career∗. Criminology, 25(2), 399–420.
https://doi.org/10.1111/j.1745-9125.1987.tb00803.x.
Krippendorf and Hayes. (2007). Answering the call for a standard reliability measure
for coding data. Communication Methods and Measures, 1(1), 77–89. https://doi.
org/10.1080/19312450709336664.
Kruglanski, A. W., Gelfand, M. J., Bélanger, J. J., Sheveland, A., Hetiarachchi, M., & Gunaratna, R.
(2014). The psychology of radicalization and deradicalization: How significance quest impacts
violent extremism. Political Psychology, 35, 69–93. https://doi.org/10.1111/pops.12163.
LaFree, G., & Dugan, L. (2015). How has criminology contributed to the study of terrorism since
9/11? In Terrorism and counterterrorism today (Vol. 20, pp. 1–23). London: Emerald Group
Publishing Ltd.
LaFree, G., & Freilich, J. D. (Eds.). (2017). The handbook of the criminology of terrorism. Wiley-Blackwell. Retrieved from https://www.wiley.com/en-us/
The+Handbook+of+the+Criminology+of+Terrorism-p-9781118923955
LaFree, G., Jensen, M. A., James, P. A., & Safer-Lichtenstein, A. (2018). Correlates of violent political extremism in the United States. Criminology, 56(2), 1–36. https://doi.
org/10.1111/1745-9125.12169.
Laub, J. H., & Sampson, R. J. (1993). Turning points in the life course: Why change matters to
the study of crime. Criminology, 31(3), 301–325. https://doi.org/10.1111/j.1745-9125.1993.
tb01132.x.
Loeber, R., & Le Blanc, M. (1990). Toward a developmental criminology. Crime and Justice, 12,
375–473. Retrieved from http://www.jstor.org/stable/1147443.
Luoma, J. B., Twohig, M. P., Waltz, T., Hayes, S. C., Roget, N., Padilla, M., & Fisher, G. (2007).
An investigation of stigma in individuals receiving treatment for substance abuse. Addictive
Behaviors, 32(7), 1331–1346. https://doi.org/10.1016/j.addbeh.2006.09.008.
The Link Between Prior Criminal Record and Violent Political Extremism in the United… 145
Lussier, P., Proulx, J., & Leblanc, M. (2005). Criminal propensity, deviant sexual interests and
criminal activity of sexual aggressors against women: A comparison of explanatory models∗.
Criminology, 43(1), 249–282. https://doi.org/10.1111/j.0011-1348.2005.00008.x.
Manza, J., & Uggen, C. (2008). Locked out: Felon disenfranchisement and American democracy
(1st ed.). New York: Oxford University Press.
Matsueda, R. L. (1992). Reflected appraisals, parental labeling, and delinquency: Specifying a
symbolic interactionist theory. American Journal of Sociology, 97(6), 1577–1611. https://doi.
org/10.1086/229940.
McCauley, C., & Moskalenko, S. (2011). Friction: How radicalization happens to them and us.
Oxford: Oxford University Press.
McGloin, J. M., Sullivan, C. J., Piquero, A. R., & Pratt, T. C. (2007). Local life circumstances and offending specialization/versatility: Comparing opportunity and propensity models. Journal of Research in Crime and Delinquency, 44(3), 321–346. https://doi.
org/10.1177/0022427807302664.
Monahan, J., & Skeem, J. L. (2015). Risk assessment in criminal sentencing. Virginia Public
Law and Legal Theory Research Paper, 53. Retrieved from https://papers.ssrn.com/sol3/papers.
cfm?abstract_id=2662082.
Neumann, P. R. (2013). The trouble with radicalization. International Affairs, 89(4), 873–893.
https://doi.org/10.1111/1468-2346.12049.
Pager, D. (2003). The mark of a criminal record. American Journal of Sociology, 108(5), 937–975.
https://doi.org/10.1086/374403.
Picarelli, J. T. (2006). The turbulent nexus of transnational organised crime and terrorism:
A theory of malevolent international relations. Global Crime, 7(1), 1–24. https://doi.
org/10.1080/17440570600650125.
Picarelli, J. T. (2012). Osama bin Corleone? Vito the jackal? Framing threat convergence through
an examination of transnational organized crime and international terrorism. Terrorism and
Political Violence, 24(2), 180–198. https://doi.org/10.1080/09546553.2011.648349.
Piquero, A. (2000). Frequency, specialization, and violence in offending careers. Journal of
Research in Crime and Delinquency, 37(4), 392–418. https://doi.org/10.1177/002242780003
7004003.
Piquero, A. R., Farrington, D. P., & Blumstein, A. (2003). The criminal career paradigm. In Crime
and justice: A review of research. Chicago: University Of Chicago Press.
Pyrooz, D., LaFree, G., Decker, S., & James, P. A. (2017). Cut from the same cloth? A comparative study of gang members and domestic extremists in the United States. Justice Quarterly, 1,
1–32. https://doi.org/10.1080/07418825.2017.1311357.
Robins, L. N., & Przybeck, T. R. (1985). Age of onset of drug use as a factor in drug and other
disorders. NIDA Res Mongr, 56, 178–192. https://doi.org/10.1111/j.1360-0443.2010.02902.x.
Rosen, D. M. (2005). Armies of the young: Child soldiers in war and terrorism. New Brunswick:
Rutgers University Press.
Rubin, D. B. (2004). Multiple imputation for nonresponse in surveys. Hoboken: Wiley-Interscience.
Safer-Lichtenstein, A., LaFree, G., & Loughran, T. A. (2017). Studying terrorism empirically:
What we know about what we don’t know. Journal of Contemporary Criminal Justice, 33(3),
273–291. https://doi.org/10.1177/1043986217697873.
Sampson, R. J., & Laub, J. H. (1997). A life-course theory of cumulative disadvantage and the
stability of delinquency. In Developmental theories of crime and delinquency (Advances in
Criminological Theory) (Vol. 7, pp. 133–161). New Brunswick: Transaction.
Sanderson, T. M. (2004). Transnational terror and organized crime: Blurring the lines. SAIS Review
of International Affairs, 24(1), 49–61. https://doi.org/10.1353/sais.2004.0020.
Schnittker, J., & John, A. (2007). Enduring stigma: The long-term effects of incarceration on health. Journal of Health and Social Behavior, 48(2), 115–130. https://doi.
org/10.1177/002214650704800202.
Shover, N. (1996). Great pretenders: Pursuits and careers of persistent thieves (1st ed.). Boulder:
Westview Press.
146
M. A. Jensen et al.
Singer, S. I. (1986). Victims of serious violence and their criminal behavior: Subcultural theory
and beyond. Violence and Victims, 1(1), 61–70. https://doi.org/10.1891/0886-6708.1.1.61.
Steffensmeier, D. J., & Ulmer, J. T. (2005). Confessions of a dying thief: Understanding criminal
careers and illegal Enterprise (1st ed.). New Brunswick: Aldine Transaction.
Stoll, M. A., & Bushway, S. D. (2008). The effect of criminal background checks on
hiring ex-offenders∗. Criminology & Public Policy, 7(3), 371–404. https://doi.
org/10.1111/j.1745-9133.2008.00516.x.
Sullivan, C. J., Mcgloin, J. M., Pratt, T. C., & Piquero, A. R. (2006). Rethinking the “norm” of
offender generality: Investigating specialization in the short-term. Criminology, 44(1), 199–
233. https://doi.org/10.1111/j.1745-9125.2006.00047.x.
Toros, H., & Mavelli, L. (2013). Terrorism, organised crime and the biopolitics of violence. Critical
Studies on Terrorism, 6(1), 73–91. https://doi.org/10.1080/17539153.2013.765701.
Uggen, C., & Manza, J. (2002). Democratic contraction? Political consequences of felon disenfranchisement in the United States. American Sociological Review, 67(6), 777–803. https://doi.
org/10.2307/3088970.
Wolfgang, M. E., Sellin, J. T., & Figlio, R. M. (1972). Delinquency in a birth cohort. Chicago:
University of Chicago Press.
Testing a Threat Model of Terrorism:
A Multi-method Study About
Socio-­Economic and Psychological
Influences on Terrorism Involvement
in the Netherlands
Vanja Ljujic, Inge Versteegt, Frank Weerman, Fabienne Thijs,
Jan-­Willem van Prooijen, Fatima el Bouk, and Steve van de Weijer
Introduction
In this chapter, we presents the results from a multi-method study in the Netherlands
into the role of socio-economic and psychological factors underlying terrorism
involvement. Building on theories and findings of previous researchers in the field,
we present a descriptive model of terrorism that categorizes distal and proximal
‘threat triggers’. In the quantitative part of the study, we analysed a combined data
set on suspects of terrorist offenses, a control sample of the general population and
a sample of general offenders. Terrorism suspects were more often lower educated,
unemployed, and previously involved in crime compared to persons from the general
V. Ljujic
Vrije Universiteit Amsterdam, Amsterdam, Netherlands
I. Versteegt
Utrecht University, Utrecht, Netherlands
F. Weerman (*)
Erasmus University Rotterdam, Rotterdam, Netherlands
NSCR (Netherlands Institute for the Study of Crime and Law Enforcement),
Amsterdam, Netherlands
e-mail: Fweerman@nscr.nl
F. Thijs · J.-W. van Prooijen
Vrije Universiteit Amsterdam & NSCR (Netherlands Institute for the Study of Crime
and Law Enforcement), Amsterdam, Netherlands
F. el Bouk
The Hague University of Applied Sciences, The Hague, Netherlands
S. van de Weijer
NSCR (Netherlands Institute for the Study of Crime and Law Enforcement),
Amsterdam, Netherlands
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_7
147
148
V. Ljujic et al.
population with the same gender and age. Relatively often, they had lost their job or
became imprisoned for another crime a year before they were charged with a terrorist offense. In the qualitative part of the study, we conducted interviews with four
detainees from terrorist units, eight detainees charged with traditional crimes (as
reference group), and 18 professional informants that had personal experience with
current and former detainees on terrorism and other offenses. The results of these
interviews suggest that among terrorist offenders, early family experiences, attachment problems, and mental health issues increase feelings of perceived threat,
which further justify violent narratives of belonging and significance.
In this chapter, we present the results from a multi-method study (part of the
PROTON collaborative project) that was aimed to investigate the role of socio-­
economic and psychological factors and processes leading to involvement in terrorism. With this contribution, we attempt to serve three main goals. First, we formulate
a ‘threat model of terrorism’ that incorporates major socio-economic, cultural and
psychological determinants of terrorism that have been proposed in the literature.
Second, we aim to present and summarize the most important findings of our two
studies and provide quantitative as well as qualitative new insights on terrorism
offenders in the Netherlands. Third, we aim to use the quantitative and qualitative
findings to evaluate the basic features of our threat model of terrorism, and to comparing terrorist suspects with the general population as well as suspects of regular
offenses.
Previous research in the Netherlands already highlighted the potential role of
individual and group level adversities and strains, relative deprivation, personal
characteristics and network dynamics (see for overviews e.g., Weggemans et al.
2014; Feddes et al. 2015; de Bie 2016). However, most of the existing research in
the Netherlands has been based on open sources, government data and/or qualitative
material retrieved from small and specific samples of respondents. There are still
important gaps in contemporary knowledge and insights on who are involved in terrorist offenses in the Netherlands and what has led them to become involved or
recruited. Only recently, studies began to employ quantitative data of larger samples
(Weenink 2015; de Bie 2016; Bergema and van San 2019). In these studies, the
majority of jihadists was found to be from the lower social strata, often failed to
acquire adequate vocational education, and was unemployed or working in unskilled
jobs (de Bie et al. 2015; Bergema and van San 2019). Previous studies also show,
however, that there is not one common characteristic for all radicalized individuals.
Researchers agree that the answer to why and how people become involved in terrorism should be sought in complex pathways towards terrorist engagement (see
also Borum 2011). Both attitudinal and behavioral aspects of terrorism are important, and influenced by a variety of proximal and distal factors, shaped by different
interpersonal and socio-cultural socializing agents, including family, peers, and religious leaders (Sageman 2004; Decker and Pyrooz 2011; Victoroff and
Kruglanski 2009).
In our contribution, we focus not only on the characteristics of those who become
involved in a terrorist ideology or network in comparison to the general population,
but also investigate differences and similarities to those who commit regular crimes.
Testing a Threat Model of Terrorism: A Multi-method Study About Socio-Economic…
149
In this way, we might better tease out what processes are specific for becoming
involved in terrorism, as opposed to the processes underlying more general involvement in developing criminal behavior.
Theoretically we will argue that the transition from radical thoughts to radical
actions occurs in conjunction with a psychological need to (restore) clarity and
positive self-image (significance) in face of perceived threat to one’s material and
immaterial wellbeing (building on Kruglanski et al. 2014). Building on past research
and literature, we will propose a descriptive threat model of terrorism that categorizes distal and proximal ‘threat triggers’ of terrorism – experiences of (perceived)
threat that sets in motion cognitive and emotional responses, and as a result of that,
increases involvement in extremism and radical action. This model combines psychological and criminological notions of perceived threat (Stephan and Stephan
2013; Schmid and Muldoon 2015); strain (Agnew 2014) and significance loss
(Kruglanski et al. 2014) to make an inventory of the factors associated to violence-­
permissive attitudes and violent action. The model also stresses the importance of
proximal triggers and personal experiences in terrorist engagement.
We will report the results of two complementary studies, one with a quantitative
approach and one with a qualitative approach. In the quantitative study, we combined and analyzed various existing data sets that were available for the complete
Dutch population of terrorism suspects since the introduction of the Crimes of
Terrorism Act in 2004. In the qualitative study (Study 2), we complemented the
quantitative analyses with an in-depth account of the motives, experiences and life
histories of persons involved in terrorist activities in the Netherlands. To achieve
this, we conducted interviews in prisons with suspects and convicts of terrorist
offenses. We conducted interviews with four detainees from terrorist units, as well
as eight detainees charged with traditional crimes (to provide a reference group),
and complemented this with interviews with 18 informants that had personal experience with current and former detainees that were incarcerated on the basis of the
Terrorist Crime Act.1
In a concluding section, we will summarize these findings and compare them to
our theoretical model to evaluate its validity.
The Socio-Political Context of the Study
According to Statistics Netherlands (2016), the Dutch ethnic majority covers 79.3%
of the total population. Most non-Western immigrants come from Turkey (2.4%),
Morocco (2.2%) and Suriname (2.1%). In terms of religious affiliation, 4.9% of
Dutch population is Muslim.2 For decades, constitutional freedom to religion and
Both quantitative and qualitative studies have been approved by the ethical board of the Faculty
of Law from the VU university in Amsterdam.
2
See: www.cbs.nl/nl-nl/publicatie/2016/51/de-religieuze-kaart-vannederland-2010-2015.
1
150
V. Ljujic et al.
multicultural policy has facilitated integration of immigrants into the Dutch society.
However, the pace of integration of immigrants in the socio-economic sphere has
lagged behind the juridical sphere (Shadid 1991). In particular, there is a considerable amount of housing and school segregation (that is, concentration of minorities
in certain neighborhoods and schools). Combined with increased political polarization, this facilitated a social distance between the host and Muslim population
(Veldhuis and Bakker 2009). Ethnic associations, growing presence of radical
imams from abroad, and experiences of discrimination and exclusion are further
obstacles for intergroup contact. Until 9/11 however Muslim extremism did not
pose an immanent threat to Dutch society.
After the assassination of Theo van Gogh in 2004, jihadi extremism was waning
for 7 years, to be revoked again in 2011 with the start of the Arab Spring (de Bie
2016).3 Political turbulences in the Middle East and, especially, the emergence of
Islamic State rekindled the emergence of jihadists’ networks in the Netherlands.
Since 2012, the risk of terrorism has been labelled as “substantial” by the Dutch
Security Service (Bergema and van San 2019). In recent years, around 250 people
from the Netherlands have travelled to Syria and Iraq to join Islamic State (ibid.).
These foreign fighters are regarded as a major security concern in the Netherlands
(Bakker and de Bont 2016; Wittendorp et al. 2017). After the military defeat of
Islamic State, some of the combat-trained jihadi travellers returned to the
Netherlands, increasing the perceived risk of terrorist attacks in the country.
In order to counter the security threats, the Netherlands amended several legislative measures to supplement the Crimes of Terrorism Act. In our study, we adopt the
Dutch legal definition of terrorism, which explicitly refers to criminal offences
‘committed with a terrorist intent’. According to Article 83a, terrorism entails a
wide range of activities aimed at serious intimidation of the population (or part of
it); the attempts to force the government or any organization to act or refrain from
acting, and the disruption of the political, economic or social structure of a country.
Several laws, such as The Code of Criminal Procedure and the Penal Code have
been amended to investigate and prosecute crimes with terrorist intent. A list of
these crimes has been widened to include various offences, such as recruitment for
jihad; membership in a terrorist organization; the preparation of (and conspiring to)
commit terrorist offence; spreading terrorist propaganda, and providing material
and logistic support to terrorist groups. In an attempt to minimize prison radicalization and recruitment, the Dutch Ministry of Justice established special prison units
for terrorism detainees.
The salience of terrorist threat (e.g. in media and political statements) adds its
share to intergroup anxiety and tensions along ethnic and religious lines (Andersen
and Mayerl 2018; Choma et al. 2015). PEW Research Centre (2016) found that 58%
of surveyed Dutch believe that Muslims support extremist groups like ISIS and 35%
In the aftermath of the Theo van Gogh murder, two prominent jihadi networks, known as the
Hofstad group and the Context group, have been dismantled by the security services.
3
Testing a Threat Model of Terrorism: A Multi-method Study About Socio-Economic…
151
reported overly negative attitudes towards Muslims.4 A survey from 20175 shows
that 67% of Dutch population views ISIS as a major threat to the society. According
to study by Statistics Netherlands (2017), 70% of Dutch citizens are worried about
a terrorist attack, and 40% believe there is a “good chance” it may happen in the
country.6 In addition to perceived threats to physical safety, recent polls indicate a
perceived threat to national culture and the existence of xenophobic feelings. It was
reported that 83% of the Dutch are concerned about Dutch norms and values, and
81% is concerned about immigration. Fifty percent of the respondents see non-­
Western migrants as a threat to their way of life, and almost one third of the Dutch
population reported to feel threatened by the large influx of refugees.7
Right-wing extremists exploit the threat of terrorism and negative stereotypes
against Muslims to pursue a discriminatory and – in some cases – violent
Islamophobia agenda. Since 2004, two thirds of Dutch mosques have received
threats or were vandalized by hate graffiti, smashed windows, dead pigs on their
doorsteps, or even arson. These incidents indicate that Muslims as a religious group
have been held accountable for terrorist attacks performed by militant extremists
(Vellenga 2018). Right-wing extremists suggest through social media and militant
public marches that it is time to defend “our way of life” violently against the rise
of Islam. At the same time, an increasing number of non-Western (Muslim) minorities feel more insecure and less at home in the Netherlands (cf. Visser-Vogel et al.
2018). According to the Social and Cultural Planning Bureau, more than 50%
Moroccan-Dutch and 40% Turkish-Dutch have accounted discrimination in different areas of life, particularly in the job market (Andriessen et al. 2014).
Additionally, there are also left-wing extremist networks active in the
Netherlands.8 They sometimes confront right-wing extremists in public and violent
anti-demonstrations. Left-wing extremist groups are less organized, and less known
for attacks on citizens but may also pose a threat to society, as manifested in the
murder of right-wing politician Pim Fortuyn by a left-wing extremist in 2002.
However, left-wing extremists have not been terrorism suspects since the Crimes of
Terrorism Act in 2004. Therefore, they are not part of our research population.
While only a small portion of Dutch Muslim gets involved in activities with terrorist intent, many (deprived) individuals from the general population may be susceptible to radical beliefs (Balen and van den Bos 2017; Verkuyten 2018). According
to the Dutch security agency, there are a few thousand sympathizers of jihadi ideology in the country, which is attributed to ongoing polarization among majority
Dutch citizens and their Muslim counterparts (AIVD 2016).9
http://www.pewglobal.org/2016/07/11/methodology-27/
http://www.pewglobal.org/2017/08/01/globally-people-point-to-isis-and-climate-changeas-leading-security-threats/
6
https://www.counterextremism.com/countries/netherlands
7
https://www.ipsos.com/en/global-trends-survey-2017
8
See Van Ham et al. (2018), https://www.wodc.nl/binaries/2867_Summary_tcm28-323558.pdf
9
See
https://www.aivd.nl/binaries/aivd_nl/documenten/jaarverslagen/2016/04/21/jaarverslagaivd-2015/jaarverslag-aivd-2015-definitieve-webversie.pdf.
4
5
152
V. Ljujic et al.
revious Insights on Socio-Economic Influences
P
and Psychological Processes Related to Terrorism Involvement
Literature on terrorism proposes a variety of historical, political, religious, economic, socio-economic and psychological drivers of terrorism (see e.g., Schmid
2004, 2011). Most studies postulate that terrorist involvement occurs in conjunction
with social disadvantage and certain psychological responses to deal with collective
and individual strains (cf. Agnew 2016; Kruglanski et al. 2014; Doosje et al. 2013).
Research in the Netherlands shows that marginalization and socio-economic
deprivation are relatively prevalent among Dutch jihadi extremists and foreign
fighters. Some disadvantageous neighborhoods have become a breeding ground for
suburban terrorists (cf. Ljujic et al. 2017). Several authors (e.g. de Bie 2016; de Poot
et al. 2011) argue that the interplay of unfavorable life circumstances and socio-­
cultural factors determines both recruitment into, and bonding within, extremist
groups. The majority of Jihadi travellers (65.5%) in the Dutch sample were friends
with fellow fighters, often from the same neighborhood, before travelling to Syria
(Bergema and van San 2019). Unemployment rates are relatively high among jihadi
fighters; moreover, many of them are from poor backgrounds, have relatively low
levels of education, and have dropped out of secondary or tertiary school (Feddes
et al. 2015; de Bie 2016; Bergema and van San 2019; Weenink 2015).
Psychological research suggests that disruptive life events may trigger extremist
behavior (violence) due to “significance loss,” promoting a perception of life as
meaningless (Kruglanski et al. 2014). In other words, the ultimate motivation for
individual involvement in terrorism lays in a personal motivation of “significance
quest” (the search for esteem, achievement, meaning, competence, and control)
against the background of disturbing and threatening life events (cf. Kruglanski
et al. 2017). Empirical terrorism studies distinguished several ‘significance loss’
triggers for radicalization, such as the death of a loved one, failure at work or in
education, and contacts with the justice system (Feddes, Nickolson and Doosje
2015). These disruptive events may have provided a ‘cognitive opening’ for alternative life goals in the form of radical beliefs and extremist violence.
The social milieu from which terrorists are recruited is constantly changing in
line with different needs and goals of terrorist groups (Gallagher 2014). Basra
and Neumann (2016) have shown that ISIS-related radicals are typically recruited
among suburban criminal networks. Likewise, several Dutch terrorism suspects
and foreign fighters have been involved in various violent and property crimes
prior to radicalization (de Bie 2016; Ljujic et al. 2017; Basra and Neumann 2016).
This confirms the so-called crime-terror nexus described in previous literature
(Hutchinson and O’malley 2007; Makarenko 2012, see also Ljujic et al. 2017).
Several processes are suggested to explain this nexus (Basra and Neumann 2016).
Terrorist networks may turn to criminal organizations for instrumental reasons.
But offenders can also be actively recruited to become terrorists, as a way to be
redeemed by God from previous criminal acts and sins (Basra and Neumann
2016). It is also possible that both criminal and terrorist involvement are ways to
Testing a Threat Model of Terrorism: A Multi-method Study About Socio-Economic…
153
cope with the adversities of socio-economic deprivation and adversity; as a ‘suburban response to strains’ (McCauley and Moskalenko 2008).
Next to socio-economic and personal deprivation and strains, perceived threats
to one’s culture, status and physical wellbeing is mentioned as one of the most
important influences on radicalization and terrorism (cf. Huddy et al. 2007; Koomen
and van der Pligt 2009). Perceived threat does not always reflect real harm, but
rather the anticipation thereof (Stephan and Stephan 2005). For example, people
who have a lower socio-economic status or belong to a devalued minority group
typically associate higher status with hostility, repression and domination (Glick
2002), whereas people who occupy a higher status may associate lower and devalued status with (undesirable distribution of) social benefits, immoral behavior and
criminality (Fiske et al. 2002). Relationships between perceived threat and group
violence are also well documented (e.g. Ranstorp 1996; Kakar 1996), including terrorism (cf. Koomen and van der Pligt 2009). Even a symbolic (nonmaterial) threat
may determine how people think, feel and behave towards others (Feldman and
Stenner 1997; Brewer 1999; Riek et al. 2006, for meta-analyses). In the past decade,
ridiculing religious symbols and cultural values (Dutch film Fitna; Mohammed cartoons; Charlie Hebdo) have led to public outrage, death threats, even terrorist attack
against satirical journalists.
Several theories provide explanations for the decisive role of perceived threat in
modern society. According to evolutionary psychologists, being perceptive of threat
is part of a human survival mechanism (cf. Buss 2015). Fears and phobias towards
‘others’ resemble instinctive reactions to threats to survival in an ancestral environment (e.g., snakes, predators, diseases) (Bracha 2004; Neuberg and Cottrell 2006;
Schmitt and Pilcher 2004). This evolutionary ‘threat management system’ enables
people to quickly recognize threatening stimuli in their environment, including
pathogens (i.e., the disease avoidance system) and threats to one’s physical integrity
(Neuberg et al. 2011). Such physical threats include hostile coalitions or outgroups.
Indeed, Bowles (2009) illuminated that intergroup conflict and warfare significantly
shaped the process of natural selection among ancestral humans. This facilitated not
only within-group cooperation but also psychological adaptations to compete for
scarce resources (e.g. water, food, housing) and maintain group characteristics that
promote bonding and group cohesion (e.g. rituals, language, culture).
Social psychologists argue that group categorization evokes the salience of group
labels through discursive reinforcement of old-established fears and dislikes (Cuddy
et al. 2011). Such stereotypical evaluations of ‘outsiders’ reflect different types of
(perceived) threats and correspond to different levels of hostility towards different
(religious, ethnic, national) groups. Historically, perceived threat has captured stereotypical dangers from ‘others’ that were predominantly defined in the line with
perceived biological or physical differences (e.g. racism). According to the classical
theory of intergroup conflict (Tajfel and Turner 1979), people exaggerate similarities within groups (in-group cohesion) and differences between groups (Li and
Brewer 2004). Unsurprisingly, such polarization coincides with the emergence or
invigoration of fear and aggression towards ‘threatening’ outgroups (Brewer and
Campbell 1976).
154
V. Ljujic et al.
A Threat Model of Terrorism
All in all, we propose that perceived group threat, associated with relative deprivation, socio-economic hardship and personal uncertainty, increase susceptibility to
extremist viewpoints and ultimately, involvement in terrorist activities (Doosje et al.
2016; Kruglanski et al. 2014; Feddes et al. 2015; Agnew 2016; Brewer 1999).
Our model (presented in Fig. 1) postulates that perceived threat originates from
different distal and proximal sources. Sometimes, the source of threat involves factors which are far beyond the personal zone of influence. For example, people perceive distal global affairs (perceived Western dominance in geopolitical distribution
of power and wealth), and international events (military interventions against Iraq,
Syrian war, Gaza conflicts, etc.) as unjust, threatening and harmful. These distal
threats may lead to feelings of solidarity, empathy, but also anger, frustration and
grievances along lines of affiliation, such as ethnic and religious backgrounds.
We also propose discrimination and segregation as a source of threat. Consistent
to Sageman’s (2008) emergent model of terrorist involvement, we argue that resentments associated to devalued group status resonate with extremism as a device to
regain positive group identity. It involves uncritical acceptance of in-group’s rules
and values; and a willingness to defend the group ‘in threatening situations’
(Moghaddam 2005). Perceived threat also contributes to a process of dehumanization of outgroup members, and the rejection of mainstream norms and laws (van
Prooijen et al. 2015; Zick et al. 2008).
We expect that perceived threat associated to one’s personal zone (disruptive and
traumatic experiences and individual strains and adversities) plays a crucial role in
the step towards terrorist engagement. As previously mentioned, a person does not
need to encounter an imminent threat to life to feel genuinely threatened. A threat to
one’s status may have the same profound effects, because it threatens a deep-rooted
desire to matter, to be significant. According to Kruglanski et al. (2014, 2017), such
Threat triggers
Cognitive and
emotional response
Grievances,
conformational bias
Behavioral response
Distal
international
factors
Military
interventions,
ethnic conflicts
abroad
Distal
national
factors
Discrimination,
relative
deprivation,
segregation
Resentment, ‘ingroup love and
outgroup hate’
Separation from
mainstream society,
breaking rules, criminality
Proximal
personal
factors
Socio-economic
status, disruptive
experiences
Loss of significance,
violence permissive
attitudes
Crimes with terrorist intent
Fig. 1 Threat model of terrorism
Seeking contact with likeminded persons, political
activism, humanitarian
work
Testing a Threat Model of Terrorism: A Multi-method Study About Socio-Economic…
155
threats may lead to action when someone gets into contact with persons or means
(i.e. violence) to combat the source of threat (see also Doosje et al. 2013; King and
Taylor 2011; Klausen 2015). In extreme cases of suicide terrorism, perceived threats
are so immanent that they require an ultimate sacrifice in order to achieve significance again (de Graaf 2017; cf. Versteegt et al. 2018).
Study 1: Socio-economic determinants of terrorism
In our quantitative study, we utilized two data sources. An anonymized database
from the Dutch prosecution office was used to identify all suspects who were
accused of ‘crimes with terrorist intent’ in the Netherlands (based on the Terrorist
Act 2004) between 2004 and April 2017. This comprises the complete population of
terrorism suspects since the introduction of the Crimes of Terrorism Act. This dataset could be combined with several anonymous datasets on general statistics from
Statistics Netherlands (CBS) to identify quantifiable and registered indicators of
factors related to terrorism involvement. The unique combination of these data
enabled us to acquire insight into the relationship between various demographics,
socio-economic indicators of adversity, previous crime involvement and being a
suspect of terrorist activities.
We did not receive the data directly. Instead, the identifiable data about terrorism
suspects (population database numbers) from the Dutch prosecution office were
sent directly to Statistics Netherlands, where the population database numbers were
removed so that the suspects were not identifiable. The data preparation and analyses were conducted via a protected server of Statistics Netherlands.
In total, the database from the Dutch prosecution office consisted of 353 suspects
of terrorist offenses. However, a substantial portion of these persons could not be
linked to the databases of Statistic Netherlands, because they were not registered in
the Personal Records Database (BRP).10 Individuals who are not registered are most
likely illegal immigrants or asylum-seekers whose applications have been rejected.11 They constituted 21.0% of the sample. For 279 suspects, we had sufficient
information about demographics, socio-economic position and their criminal past.
This ‘terrorist’ sample (n = 279) is the basis of our quantitative analysis.
We also constructed two control groups. For both the descriptive statistics as well
as logistic regression analyses, we created a control offender sample (n = 279) that
consisted of non-terrorism suspects of other offenses who are similar to terrorism
The database contains the personal data of people who (used to) live in Netherlands. Municipalities
record the personal data of all residents in the BRP. For information, see: https://www.government.
nl/topics/personal-data/personal-records-database-brp
11
Research on Dutch jihadi networks demonstrates the involvement of vulnerable immigrants in
the Netherlands (De Bie et al. 2014). For additional information on who is excluded from the BRP,
see the CBS-website: https://www.cbs.nl/nl-nl/achtergrond/2017/38/basisregistratie-personen
(this information is only available in Dutch).
10
156
V. Ljujic et al.
Table 1 Basic demographics of terrorism suspects
Factors
Sex
Age at time of terrorist offense
Category
Male
Female
17 or younger
18–25
26–35
36–45
46 or older
Terrorism suspects (n = 279)
%
#
87.5
244
12.5
35
7.9
22
32.6
91
30.5
85
19.7
55
9.3
26
suspects along the demographics of age and gender.12 Also a general control sample
(n = 6000) of citizens from the general population was selected, again similar with
regard to age and gender.
The datasets from Statistic Netherlands gave us access to the necessary data.
However, most of the datasets, which we combined with the dataset from the Dutch
prosecution office, did not offer information on the complete time span (2003–2016)
that we focused on. Therefore, the tables below show different sample sizes for different variables. It also appeared that there was a lot of missing data on education in
general. For this reason, we chose to exclude the education variable from the logistic regression analyses.
Finally, for the dynamic variables that may change over time (e.g., work situation,
level of education, suspected of other crimes, et cetera) we used data from the year
before the suspicion.13 This way we ensured that potential relations we find cannot
be interpreted as effects of being a terrorism suspect instead of a potential cause.
In Table 1, we present some general demographic descriptives for the total group
of 279 suspects of terrorist offenses for which we had sufficient information. Males
dominate within the terrorism suspect population (87.5%). The mean age is 30
(median: 27) during the time of suspicion and the majority of the population is
between 18 and 35 years of age (62.6%).
In Table 2, we describe the characteristics of our sample of terrorism suspects
together with our two control samples of general suspects and the general population. We present information about origin, level of education, whether they were
officially employed or not, and previous involvement in general offending.
The table shows that suspects of terrorist activities have a migrant background
more often than general suspects and the general population (81.3% versus 48.4%
and 33.2%). This may be the result of the recent wave of jihadi terrorist activities but
also reflects the political and police attention that is devoted to this type of terrorism
at the moment. Further, suspects of terrorist offenses seem slightly less often
­officially employed (44.0% versus 50.9% and 58.7%), and they more often lost their
For age, we made age categories and matched the age groups of the terrorism suspects with the
control groups.
13
For the dynamic variables of the control groups (which were based on 2016 information), we
used data from the year 2015.
12
Testing a Threat Model of Terrorism: A Multi-method Study About Socio-Economic…
157
Table 2 Origin, SES and criminality of terrorism suspects, general suspects and population
Factors
Origin
Category
Non-­
immigrant
First
generation
Second
generation
Terrorism
suspects
(n = 279)
%
#
18.6
52
General
suspects
(n = 279)
%
#
51.6 144
General
population
(n = 6000)
%
#
66.8 4010
41.2
115
21.9 61
19.7 1181
40.1
112
26.5 74
13.5 809
(n = 170)
106
57
7
(n = 250)
110
140
(n = 250)
32
12
206
(n = 224)
(n = 4235)
67.4 151
45.2 1915
28.1 63
36.9 1563
4.5 10
17.9 757
Highest level of education
Lower
Middle
Higher
62.4
33.5
4.1
Yes
No
44.0
56.0
Employmenta
Job loss?
(A year before the
suspicion compared to
2 years before the
suspicion)
Suspected of offense(s)
Job loss
11.5
Found a job 4.3
No change 84.2
Yes
No
29.2
70.8
Yes
No
9.9
90.1
Detention
50.9 142
49.1 137
58.7 3519
41.3 2481
6.1 17
9.0 25
84.9 237
3.5 208
6.5 389
90.1 5403
(n = 233)
68
29.4 82
165
70.6 197
(n = 233)
23
7.2 20
210
92.8 259
2.4 146
97.6 5854
0.5 31
99.5 5969
The employment variable exists of data on whether or not a person had a job in a specific year.
Those without a job can be unemployed on a voluntary or involuntary basis. Therefore, besides
being unemployed, those who fall within the category ‘No’ could also be students (and have no
need for a (official) job), live from social benefits, or have unofficial jobs
a
job in the year before the suspicion (11.5% versus 6.1% and 3.5%). Later in the
analyses we will scrutinize whether these differences also point to significant and
independent effects of employment variables. Education levels of terrorist and general suspects are quite similar; in both groups, lower education makes up (almost)
two third of the group. Within the general population, less than half of the sample
falls within the lower education category. A substantial portion of the terrorism
suspects, almost 30%, was involved in general types of offending before they
became suspected of terrorist offenses. This is similar to the general suspects, but
highly different from the general population of which almost none were suspected
of offenses (2.4%). Finally, only a minority has spent time in detention 1 year before
becoming a terrorism suspect (9.9%). However, this is comparable to the general
158
V. Ljujic et al.
Table 3 Associations between being a terrorism suspect and origin, employment and previous
offending variables
Factors
Origin
Non-immigrant
First generation
immigrant
Second generation
immigrant
Employment
Job loss
Found job
Suspected of criminal
offenses
Detention
∗
Correlation coefficient Phi
Terrorism suspect vs. general
population
Terrorism suspect vs. general
offenders
−0.208∗∗∗
0.110∗∗∗
−0.345∗∗∗
0.208∗∗∗
0.155∗∗∗
−0.058∗∗∗
0.086∗∗∗
−0.018
0.273∗∗∗
0.191∗∗∗
0.144∗∗
−0.069
0.095∗
−0.094∗
−0.002
0.049
p < .05. ∗∗p < .01. ∗∗∗p < .001
suspects (7.2%), and much higher than in the general population, were we find that
almost no one in the sample (0.5%) was detained before.
In the next analysis, we investigated to which extent the variables from Table 2
had an independent and significant relationship with being a suspect of terrorism.14
We included 204 terrorism suspects in this analysis for whom we had sufficient
information.15
Table 3 gives an overview of the correlations between different factors and being
suspected of a terrorist offense. In the comparison with the general population, we
mostly find values between 0.1 (small) and 0.3 (medium) and all of the correlations
are significant. We observe a modest positive relation between being a terrorism
suspect and having a 1st and 2nd generation immigrant background, being suspected
of criminal offenses, and having been in detention. For non-immigrant background
we find a negative correlation and for employment (having a job) a very small negative one. The comparison with general offenders, on the other hand, demonstrates
that solely origin is significantly correlated with being a terrorism suspect instead of
suspects of general types of crime. Here, the other variables are very weakly and
non-significant associated with being a terrorism suspect instead of a general suspect.
Table 4 present the results from two logistic regression models that give us
insight in what characteristics/factors are independently associated with an
increased likelihood of becoming a terrorism suspect. In regression Model I, terrorism ­suspects compared to the general population, most of the effects are signifiAs explained before, the education data was excluded from the logistic regression analyses due
to the structural missings.
15
For some terrorism suspects, we had missing data on a range of variables. Therefore, we excluded
them from the analyses.
14
Testing a Threat Model of Terrorism: A Multi-method Study About Socio-Economic…
159
Table 4 Summary of logistic regression analyses for variables predicting being a terrorism suspect
Variable
Origin
Non-immigrant
1st generation
immigrant
2nd generation
immigrant
Job loss (trigger)
Found job
(trigger)
No official
employment
Suspected of
criminal offenses
Detention
N
Model fit
Model I: Terrorism suspect vs. general Model II: Terrorism suspect vs.
population
general suspects
Adj.
Adj.
odds
odds
ratio 95% CI OR
ratio 95% CI OR B (SE)
B (SE)
(ref.)
1.795 (.195)∗∗∗ 6.02
[4.11, 8.82]
(ref.)
1.417 (.243)∗∗∗ 4.13 [2.56, 6.64]
1.891 (.203)∗∗∗ 6.62
[4.45, 9.86]
1.157 (.239)∗∗∗ 3.18 [1.99, 5.08]
1.195 (.255)∗∗∗ 3.30
−.731 (.380)
.48
[2.01, 5.44]
[.23, 1.01]
.507 (.356)
−.657 (.435)
1.66 [.83, 3.33]
.52 [.22, 1.22]
.143 (.216)
1.15 [.76, 1.76]
2.413 (.203)∗∗∗ 11.17 [7.50, 16.62] −.067 (.225)
1.236 (.362)∗∗ 3.441 [1.69, 7.00]
6204
R2 = .228 (Nagelkerke)
Hosmer and Lemeshow test: .455
.94
[.60, 1.45]
.343 (.356)
1.41 [.70, 2.83]
558
R2 = .144 (Nagelkerke)
Hosmer and Lemeshow test: .283
∗
p < .05. ∗∗p < .01. ∗∗∗p < .001
cant.16 People with an immigrant background have a higher probability of becoming
a terrorism suspect than others in the population. Also losing a job in the previous
year predicts an increased likelihood of becoming a terrorism suspect. Interestingly,
the opposite situation of finding a job does not have an effect. Also previously being
suspected of other criminal offenses is related to an increased likelihood of becoming a terrorism suspect, and the effect is relatively strong (estimated odds ratio of
more than 11). Finally, being in detention in the year before is also positively related
to an increased likelihood of becoming a terrorism suspect.
Model II, terrorism suspects compared to general suspects, shows that none of
the effects related to work situation or criminality are significantly related to the
distinction between being a terrorism suspect and a general suspect. Solely origin,
having an immigrant background compared to not having an immigrant background,
predicts an increased likelihood of becoming a terrorism suspect instead of a general suspect. In line with the descriptive data, this confirms that terrorism suspects
and general suspects are similar on most of the characteristics discussed.
Inclusion of the variable unemployment in Model I resulted in a significant Hosmer en Lemeshow
test, which is an indication that the model is not a good fit. Therefore, we ran Model I without this
variable.
16
160
V. Ljujic et al.
In sum, origin (1st and 2nd generation immigrant), job loss, being suspected of
other offenses and detention can be seen as risk factors in getting involved in terrorist crimes. However, terrorism suspects do not significantly differ from general suspects on SES characteristics and criminal history.
tudy 2: Qualitative Analysis of Perceived Threat
S
and Individual Involvement in Terrorism
In our qualitative Study 2 (for an extensive report, see Versteegt et al. 2018), we
interviewed 30 people to capture the life events, socio-economic backgrounds and
involvement mechanisms of people who were detained as a suspect of terrorist
offenses. We conducted our study in the two prisons in The Netherlands that have a
so-called terrorist unit.
We were able to recruit four inmates in the terrorism units, which we interviewed
extensively. To contextualize our findings about the terrorism suspects, we interviewed eight inmates of other units of the prison. Because the participation among
the detainees of the terrorist unit was low, we also interviewed 18 persons that were
working in or around the terrorist units and had in-depth knowledge about current
and former detainees that were suspected of terrorist offenses. These ‘informants’
included prison wardens and mentors, spiritual caretakers and psychologists at the
prison, which ensured a variety of perspectives. We also conducted a focus-group
interview with informants inside the prison.
We used a semi-structured questionnaire to assess the background, socialization
and previous development of the detainees, as well as their ideology and experiences with socio-economic and personal setbacks and adversities. Additionally, we
employed a calendar instrument, adapted from a version that was developed in a
previous study on the lives of incarcerated women (Joosen and Slotboom 2015).
With this instrument, detainees could indicate the socio-economic situation and
events in their personal life during the last year prior to their offense. We also asked
the informants to complete a life event calendar for one particular detainee that they
had sufficient information on. Through various open questions, we looked for signals of socio-economic and personal adversities, we investigated their psychosocial
wellbeing and mental health, and asked about their religious and ideological attitudes. We also asked details about their family background, childhood experiences
and socialization.
The use of informants may seem limited because they can have instrumental
reasons to express particular views about the inmates of the terrorist units. However,
we have taken several measures to increase and evaluate the reliability of the informant reports. In order to prevent a biased response from only prison workers, we
interviewed lawyers as well, and people working for the probation office – both
professionals with their own point of view, and not employed by the prison.
Informants from these different groups were generally in agreement about the ter-
Testing a Threat Model of Terrorism: A Multi-method Study About Socio-Economic…
161
rorism suspects. Further, it turned out that two of the calendars taken from informants referred to an individual who were also in our sample as a respondent. This
enabled us to compare the information in the life histories from two sources, and
these appeared to be remarkably consistent. Therefore, we do not believe that the
responses from the informants are systematically biased. Workers in the prison did
express criticism of the prison system and specific legal procedures associated with
the terrorist unit though, but we collected these statements anonymously in a separate document for the prison director and left this out of the analysis.
To guarantee anonymity, we are very careful in how we present our data. As the
terrorist units house less than 30 prisoners in total, and the professionals working
with them are limited in number as well, we present the quotes without specific
reference to age, gender or profession. In our presentation of the interview quotes,
the inmates are referred to as “detainees”. The various people with professional
personal contact with the terrorist suspects are referred to as “informants”.
We provided all the respondents with a distinguishing code for their category,
and a number for each individual. These are the codes that we used to present
the quotes:
A: terrorist suspects
B: suspects of regular criminal offences (the control group)
C: professionals working inside the prison
D: professionals working outside the prison
Our findings partly confirm the quantitative findings, but also nuance these and add
some other elements that may be important to understand why and how people
become involved in terrorism.
With regard to education, there appeared to be a difference between detainees
from the terrorist units that were regarded as ‘leaders’ and detainees that were seen
as merely ‘followers’. The followers had a relatively low level of education, while
the average educational level of the leaders was higher than that of the followers and
also the regular detainees. However, the informants reported that in many cases,
their acquired status and their economic situation did not match their educational
level and their capacities. This difference between expectations and real outcomes
may have contributed to feelings of significance loss in this group, leaving them
with cognitive dissonance about the contrast between their intellectual abilities and
their experienced opportunities in society. These experiences may have made them
susceptible to an extremist ideology.
Informant D10, working outside the prison, states: “One person had a very high education
and he literally said: so they keep rejecting me because of my ethnic background, I probably
don’t belong. Let me find a place where I do belong and where they do respect me.”
Similarly, informant C13, working in the prison, notes:
They (the whole group, IV) are rather well educated; they are not doing so badly. There are
many people in a similar situation, who do not radicalize. (Later): In their educational situation, they are not the needy. Very often, they have a better than average intelligence and
they have read many books, too.
162
V. Ljujic et al.
With regard to employment, the interviews provide additional information to the
quantitative study that could not distinguish between jobless people, students, and
under aged persons. . In our qualitative study, we found that previous employment
through jobs and independent companies is not uncommon among the interviewed
suspects at the terrorism unit, while long term unemployment was more common in
the group of regular detainees that we interviewed (4 of 8 cases). The informants
confirmed that unemployment was not the rule for terrorist suspects.
Interviewer: Did they have jobs?
Informant C3: Definitely.
I: More than at the regular prison?
C3: Oh yes, much more.
The jobs of the terrorism suspects varied from simple technical jobs at a low wage,
to management jobs. The general impression of the informants was that suspects at
the terrorist units had better jobs than the regular detainees. However, in several
cases, terrorist suspects had lost their job a few years prior to their crime. This was
reported by three of the four respondents in the terrorism unit. Two of our respondents explained how they had become obsolete in a company that they had worked
in for a longer period of time. This is what a detainee (A14) at the terrorist unit told
us about employment:
At the moment of the terrorist offense, “I was unemployed. I had been fired after a large
reorganization in a company where I had worked over 10 years, and where I had enjoyed
working. After losing his job, I unsuccessfully applied for jobs.” The terrorist suspect added
that the strained job market made it difficult to find employment.
We found no indication, however, that the income situation was extremely stressful
for the terrorism unit inmates prior to their offence. Debts, receiving social benefits
and criminality were not mentioned more by the respondents and informants (even
slightly less) for the terrorist unit than by the respondents from the control group.
With regard to housing, detainees at the terrorist unit seemed to be less deprived
than regular detainees. One of our informants describes the general picture of the
terrorist unit group in terms of housing:
Informant inside the prison, C4: “Most of them lived with their parents or with a partner.”
A similar picture of relatively stable housing conditions emerged from a focus
group interview with informants working inside the prison:
Focus group informant: “In all those years in the terrorist unit, we had only five persons
who were homeless.”
Overall, there are no outspoken objective differences between the two research
groups (terrorist unit detainees and regular detainees). We did not find striking differences in socio-economic markers like education, jobs, and housing between terrorist and regular detainees – and there are indications that the terrorist unit detainees
may even be slightly “better off”.
Next to these socio-economic variables, psychosocial and mental health factors
may have contributed to the development of extremist thought (see Weenink 2015).
Testing a Threat Model of Terrorism: A Multi-method Study About Socio-Economic…
163
However,our informants did not agree on this topic. Some of them mentioned
ADHD, post-traumatic stress, autism and attachment issues for the detainees of terrorism units. According to some informants, PTST and attachment disorders may
have made detainees vulnerable to social failure and deprivation, but they may also
have made them more susceptible to group pressure. Some informants mentioned a
proneness to violence, possibly related to psychopathology and narcissism, for the
supposed leaders. One informant explained the difference in mental illnesses and
disorders between leaders and followers as follows:
Informant C2, working inside the prison: “There are leaders and followers. Besides the
difference in IQ, these people are also different in their personalities. The leaders are natural leaders, they are highly educated and could also have been leaders in some other organization. They tend to have these personality disorders like narcissism, bipolar disorder, or
anti-social disorders. The followers have a low education and are easily influenced. They
are very vulnerable, can have depressions and a low self-image”.
In terms of adverse experiences, terrorist unit and regular detainees also appear to
share many characteristics. Both groups report various experiences of discrimination, trauma, and personal setbacks. However, the terrorist detainees seem to perceive these experiences as more severe and as signs of social injustice, while the
regular detainees seemed to accept their problems as part of life and as something
they should cope with personally.
Informant C1, working in the prison: “I always wonder if they have experienced more
losses, I don’t think so. But they are triggered by these events to question life.”
Respondent A22: “If I had to describe them (his fellow inmates at the terrorist unit): they
are sensitive guys. They cannot deal with injustice. I can let go of things. They cannot.”
These findings left us wondering why adverse events created more significance
issues, and stronger feelings of relative deprivation in the inmates from the terrorist
units. We found three possible, interrelated characteristics of the terrorist group,
which may explain these differences in perception.
First, we have found indications that the early family life of the extremist group
was less stable – this notion is based more on the accounts of our informants than
on the stories of our respondents, who seemed to provide mainly social desirable
answers about their family life. The informants mentioned that divorce of parents
and dysfunctional, broken families are very common among the detainees of the
terrorist unit. According to some of our informants, such circumstances occurred in
more than 80% of the cases (expressed in a focus group interview in prison, as well
as estimated by one informant outside prison, D10). In many families, parents were
(said to be) unable to provide stability and unconditional love. Other factors that
were mentioned included culturally transferred patterns of strict or abusive parenting, or ‘spoiling’ and pedagogical-moral neglect. Such impoverished settings may
create trauma in young children – especially when their family stands out negatively
in a larger cultural community, and when criticism of parents is a cultural taboo.
An interview with an informant, working at the terrorist unit, illustrates this:
C4: “(….) There was more instability in the situation at home, which is my impression.”
I: “Can we speak of attachment disorders?”
164
V. Ljujic et al.
C4: “Absolutely. Whereas the image of mother, within the culture, remains intact. There is
a big discrepancy there. Parents must be talked about with a lot of respect. Mother is almost
a saint. ‘Paradise is at the feet of the mother’, that is how it is written in the Qur’an. But
this particular mother often neglected things, or refrained from doing things, or was unable
to do them.”
Another illustration from a focus interview with professionals working at the terrorist unit:
Focus group informant: “In at least 90 % [other person corrects this number] or definitely
80 %... we see broken families, a criminal past, drug issues, no relationship with the father.
In such a situation it may happen that you start seeing yourself as a loser. And then suddenly, someone shows up…. They did not finish their school, they never finished anything.
Very often, youth care has been involved. We have had someone here, he kept searching. He
had been an animal activist, he had been with the Lord, and now he ended up in radical
Islam. It was just never right, and he kept looking for something else. Looking for recognition, looking for brotherhood. To belong somewhere, because they failed, that’s how they
feel.”
Second, in interviews with detainees from the terrorist units as well as with the
informants we found indications that a search for meaning, identity and belonging
was strongly present in the lives of the terrorist unit detainees. They seemed, more
so than the control group, sensitive to a narrative that provides them with an identity,
a sense of belonging, and moral guidelines from authority figures they can love as
their surrogate family.
Informant inside the prison, C4: “There is a difference. These people really have brothers.
Normal detainees have friends that buy drugs from them. These people have a brotherhood,
which really feels very intense to them.
Informant inside the prison, C1: “I can imagine that this brotherhood really means a lot to
them. When you feel you have always been treated like an outcast actually…and suddenly
there is someone who…uses these ‘lover boy’- like methods…With all those hugs, too…”
I: “Hugs?”
R: “They keep embracing each other on every occasion”.
This search for belonging may be interpreted as a result of to the reported early life
instabilities. The (mental or physical) unavailability of a primary caretaker may
have generated attachment problems in children, and as young adults, this may have
left them with an anxiety-driven need to find social bonds. The narrative of ‘brotherhood’ in extremist organizations literally provides individuals with a stable and reliable family. Persons with attachment problems also often seek identification figures
to admire. We heard stories among our informants about celebrity worship syndromes of detainees in the terrorist units (towards Mohammed B., the killer of Theo
van Gogh, for example), and religious obsessions. Also the tendency of the terrorist
unit inmates to hug each other, and form strong social pacts, may be seen as indications that the need for social bonding in these individuals has pathological and
obsessive elements, which could stem from childhood trauma.
A third factor that set apart the detainees from the terrorism unit from the regular
detainees, were the coping strategies and narratives they used to deal with adversities. We found that regular detainees more often relied on religious coping strategies
Testing a Threat Model of Terrorism: A Multi-method Study About Socio-Economic…
165
to deal with experiences of personal or social injustice or fell back on reasoning that
enabled them to keep their faith in social justice and humanity (‘Just World
narratives’).
Respondent from the control group, B2: “I found out, in life, you just have to make the best
of it. It can be over any time” (…). “You have to be able to deal with difficult situations. You
have to see through the darkness, towards the light.
Such examples of ‘coping’ narratives were not found in the stories from the terrorism unit respondents. Regular detainees were also less distrusting of the prison staff
and held compassionate views of their fellow human beings and their mistakes.
Terrorist unit detainees on the other side tended to judge all events as further contributions to their victimhood, according to multiple accounts of our informants.
Events that had a negative outcome for them were interpreted as deliberate mistreatment and discrimination, even if they probably were not.
A discussion with a focus group of informants inside the prison illustrates this:
I: “Do they talk about discrimination?”
Focus group informant 1: Oh yes. That is the main storyline.
I: The main storyline?
Focus group informant 1 and 2: Yes, they will say they are victims, and the government hates
them.
I: Do you feel that they are more than the other group using stories of discrimination and
deprivation?
Several informants: “Oh yes. All the time”
Focus group informant 1: “Anything. They return from the court, and things did not work
out positively, then everybody is plotting against them. Here, at prison, everyone is plotting
against them.
Adding to this, the inmates at the terrorist unit seem to worry a lot about social
injustice, morality, world problems, and their own destiny, according to informants.
They were described as psychologically vulnerable and less able to cope with setbacks in comparison to the regular detainees, who expressed more healthy coping
strategies.
Informant C4, working inside the prison: “They never felt able to cope with injustice. But
due to the IS ideology, they suddenly feel stronger (…).”
Lastly, we found that the role of religion (Islam) in the development of involvement
in jihadi terrorism is different than often thought. Most of the terrorist detainees
seemed to be not very well informed about the Islam and did not have a thorough
knowledge of the Quran. Instead, they became attracted to the powerful representation of the Salafi – Jihadi discourse on the internet of the Islam as “one universal
truth” in which the hypocrites must be discarded from the true believers. For them,
this resonates with the vague notions of Islam their parents have provided them
with, notably that of ‘being a good Muslim’, and it makes them believe that this
representation of Islam is correct.
Detainee A13 from the terrorist unit explains: “I had quit using alcohol, I was making my
life better. I wanted to be a good Muslim. I was looking for something new, actually I was…
looking for my identity. Then I talked with a couple of men, friends of mine, about Syria and
166
V. Ljujic et al.
they said, that is where you should go. If you want to be significant as a Muslim. There is no
war at all. There are just Muslims who need your help. So I wanted to go there, see what I
could do, make a new start.
Similarly, in the accounts of one detainee in our sample and several informants, it
appears that right-wing extremist detainees were politically not very well informed.
Instead, they were sensitive to the general narrative of ‘the foreign threat to the Dutch
identity and society’ together with a desire to help protect society from these dangers.
Conclusion and Discussion
In this chapter, we first proposed a threat model of terrorism, as a theoretical framework for organizing and interpreting the major socio-economic and psychological
determinants of terrorism involvement. Drawing upon intergroup notions of strain,
perceived threat and significance loss, our model lays out a clear distinction between
proximal and distal threat triggers, and identifies multiple emotional and cognitive
processes linking those triggers to the behavioural outcome: unlawful activity with
terrorist intent.
We utilized recent quantitative and qualitative data to test the model. In the quantitative study we combined and analysed existing data sets that were available for
the complete Dutch population of suspects of terrorist offenses since the introduction of the Crimes of Terrorism Act in 2004. The findings show that terrorism suspects belonged more often to ethnic minority groups, were lower educated, were
unemployed, and were more likely previously involved in crime, than persons from
the general population with the same gender and age. They also more often lost their
job and were detained previous to becoming involved in terrorist offenses. Among
the terrorism suspects, 30% were registered for other criminal offense(s) in the year
before becoming known to police for terrorist offense. In the multivariate analyses
we found that economic adversity and criminality substantially and independently
increase the likelihood of becoming a terrorism suspect for the general population.
These findings offer initial support the proposed threat model as well as its theoretical antecedents (Doosje et al. 2016; see also Ljujic et al. 2017; Pare and Felson
2014). However, terrorism suspects did not differ significantly from general suspects in terms of individual strains and adversities. This means that additional factors and processes are needed to fully understand why people become involved in
terrorist activities.
In the qualitative study, we complemented the quantitative analyses with an in-­
depth account of the motives, experiences and life histories of persons involved in
terrorist activities in the Netherlands. Consistent with the first study, the findings
suggest that terrorist offenders and other, regular, detainees have similar life courses
to a certain extent, and both groups are not very different in socio-economic position and experiences of deprivation and discrimination. Moreover, the qualitative
study shows that both groups also often encountered personal adversities in other
Testing a Threat Model of Terrorism: A Multi-method Study About Socio-Economic…
167
life spheres and were reported to have various types of mental health problems.
Different from general offenders, however, terrorist offenders perceive these setbacks and experiences of deprivation as more severe and signs of social injustice.
We found that this was related to three other characteristics that were more typical
for the terrorism offender group: an instable and impoverished early family life, an
intense search for meaning and identity and need for acceptance, and coping strategies that were not adaptive but instead constantly reaffirmed their perceived victimhood. In the end, these suspects have encountered ‘a bunch of guys’ that introduced
them to a network with an extremist narrative or were attracted by the strong but
simplified narratives they found on the internet. This susceptibility to ideological
(peer) indoctrination went hand in hand with their strong need for significance and
belonging, and may be explained by several psychological conditions, such as
attachment issues and mental disorders that impair social reasoning and impulse
control. The childhood problems of the terrorism suspects may have created a susceptibility to find belonging in deviant social networks. This susceptibility was further aggravated by deprivation, and experiences of discrimination, social exclusion
and loss. The narrative of the extremist networks specifically addresses these feelings of rejection and puts it in a global perspective of urgency, belonging and heroism, as well as personal gain. These processes were most outspoken for suspects
with a jihadi background, but also seemed to be applicable to right-wing extremists.
In sum, the findings support the threat model but also extend it. We found that
distal threat triggers such as conflicts abroad and socio-economic differences within
the population can contribute to emotional and cognitive responses among extremists. Muslim extremism frames group threat as indications of a devalued and discriminated minority, whereas right wing extremism exploits popular resentments
about immigrants, which threaten the ethno-national identity. Besides shaping one’s
cognitive and emotional responses towards ‘threatening’ others (Ljujic et al. 2011),
threats also provide a ‘convenient’ rationale for intergroup violence (Feldman and
Stenner 1997), including terrorism (Staub 2004). However, these extremist narratives may appeal to many, but only a few become actually involved in terrorism. The
results of our research show that proximal personal factors (such as economic hardship and disruptive life experiences), may be crucial to our understanding of involvement in terrorist offenses. Adding to this, our results suggest that early family
experiences, attachment problems, and mental health issues may have contributed
significantly to feelings of perceived threat and a susceptibility to appealing narratives that promise feelings of belonging and significance.
To a certain extent, these life histories that are associated with involvement in
terrorism and extremism may be a matter of chance and a series of bad luck, as well
as meeting the wrong persons or internet sites at the wrong time. However, our
study also provides evidence that one’s path to terrorism is driven by tangible socio-­
economic and psychological factors that might be addressed by professionals, policy makers and the criminal justice system. In this respect, Dutch society is facing a
twofold challenge: to prevent engagement in terrorism, but also, to stimulate de-­
engagement and re-socialization of convicted terrorism offenders. To these ends, a
better understanding of the role of perceived threat in radicalization trajectories (cf.
168
V. Ljujic et al.
McCauley and Moskalenko 2008) is necessary to counter extremism and terrorism
and to prevent recidivism.
It is important to note that our study has various limitations. A strong feature of
our quantitative study was that we were able to get access to data about the complete
population of terrorism suspects since terrorism was distinguished as a separate
offense in the Netherlands. However, a limitation was that we could only retrieve
general information about socio-economic characteristics and previous involvement
in crime, and could not go into detail about life histories and events. The findings
were also limited in the sense that for a part of our sample, we lacked sufficient
information on characteristics such as education and employment situation. A
strong feature of our qualitative study that we gained access to respondents and
informants in specialized prison terrorist units, that enabled us to get access to in-­
depth information about the life histories and personal backgrounds of current and
former terrorism suspects, as well as regular detainees. However, the findings were
limited in the relatively low number of respondents that we could speak, and also
had to rely on information we got from informants. Although the information from
the informants was often consistent and uniform, and were quite similar for those
working inside and outside prison, we cannot rule out the possibility that their
observations are incomplete or that experiences with some detainees in the past
were generalized too much to the complete population.
All in all, our findings offer interesting and plausible insights about the backgrounds of terrorism offenders in the Netherlands and the processes that may lead
to their involvement in terrorist crimes, but our results need to be scrutinized in
future studies to get more certainty about their validity and generalizability. We
therefore hope that the findings presented in this chapter will act as a stimulus for
further research in the field.
References
Agnew, R. (2014). General strain theory. In G. Bruinsma & D. Weisburd (Eds.), Encyclopedia of
criminology and criminal justice (pp. 1892–1900). New York: Springer.
Agnew, R. (2016). General strain theory and terrorism. In G. LaFree & J. D. Freilich (Eds.), The
handbook of the criminology of terrorism (pp. 119–132). Chichester: John Wiley & Sons.
Andersen, H., & Mayerl, J. (2018). Attitudes towards Muslims and fear of terrorism. Ethnic and
Racial Studies, 41(15), 2634–2655.
Andriessen, I., Fernee, H., & Wittebrood, K. (2014). Ervaren discriminatie in Nederland. The
Hague: Sociaal en Cultureel Planbureau.
Bakker, E., & de Bont, R. (2016). Belgian and Dutch jihadist foreign fighters (2012–2015):
Characteristics, motivations, and roles in the War in Syria and Iraq. Small Wars & Insurgencies,
27(5), 837–857.
Balen, M., & van den Bos, K. (2017). Over waargenomen onrechtvaardigheid en radicalisering.
Justitiële Verkenningen, 43(3), 31–44.
Basra, R., & Neumann, P. R. (2016). Criminal pasts, terrorist futures: European jihadists and the
new crime-terror nexus. Perspectives on Terrorism, 10(6), 25–40.
Testing a Threat Model of Terrorism: A Multi-method Study About Socio-Economic…
169
Bergema, R., & van San, M. (2019). Waves of the black banner: An exploratory study on the Dutch
jihadist foreign fighter contingent in Syria and Iraq. Studies in Conflict & Terrorism, 42(7),
636–661.
Borum, R. (2011). Radicalization into violent extremism I: A review of social science theories.
Journal of Strategic Security, 4(4), 7–36.
Bowles, S. (2009). Did warfare among ancestral hunter-gatherers affect the evolution of human
social behaviors? Science, 324(5932), 1293–1298.
Bracha, H. S. (2004). Freeze, flight, fight, fright, faint: Adaptationist perspectives on the acute
stress response spectrum. CNS Spectrums, 9(9), 679–685.
Brewer, M. B. (1999). The psychology of prejudice: Ingroup love and outgroup hate? Journal of
Social Issues, 55(3), 429–444.
Buss, D. (2015). Evolutionary psychology: The new science of the mind. New York: Psychology
Press.
Brewer, M. B., & Campbell, D. T. (1976). Ethnocentrism and intergroup attitudes: East African
evidence. Beverly Hills: Sage.
Choma, B. L., Charlesford, J. J., Dalling, L., & Smith, K. (2015). Effects of viewing 9/11 footage
on distress and Islamophobia: A temporally expanded approach. Journal of Applied Social
Psychology, 45(6), 345–354.
Cuddy, A. J., Glick, P., & Beninger, A. (2011). The dynamics of warmth and competence judgments, and their outcomes in organizations. Research in Organizational Behavior, 31, 73–98.
de Bie, J. L. (2016). How jihadist networks operate: a grounded understanding of changing
organizational structures, activities, and involvement mechanisms of jihadist networks in the
Netherlands. Leiden: Leiden University.
de Bie, J. L., de Poot, C. J., & van der Leun, J. P. (2014). Jihadi networks and the involvement
of vulnerable immigrants: Reconsidering the ideological and pragmatic value. Global Crime,
15(3–4), 275–298.
de Bie, J. L., de Poot, C. J., & van der Leun, J. P. (2015). Shifting modus operandi of jihadist foreign fighters from the Netherlands between 2000 and 2013: A crime script analysis. Terrorism
and Political Violence, 27(3), 416–440.
de Graaf, B. A. (2017). Terrorisme-en radicaliseringsstudies: Een explosief onderzoeksveld.
Justitiele Verkenningen, 43(3), 8–30.
de Poot, C. J., Sonnenschein, A., Soudijn, M. R. J., Bijen, J. G., & Verkuylen, M. W. (2011). Jihadi
terrorism in the Netherlands. The Hague: Boom Juridische Uitgevers.
Decker, S., & Pyrooz, D. (2011). Gangs, terrorism, and radicalization. Journal of Strategic
Security, 4(4), 151–166.
Doosje, B., Loseman, A., & van den Bos, K. (2013). Determinants of radicalization of Islamic
youth in the Netherlands: Personal uncertainty, perceived injustice, and perceived group threat.
Journal of Social Issues, 69(3), 586–604.
Feddes, A. R., Mann, L., & Doosje, B. (2015). Increasing self-esteem and empathy to prevent
violent radicalization: A longitudinal quantitative evaluation of a resilience training focused on
adolescents with a dual identity. Journal of Applied Social Psychology, 45(7), 400–411.
Feddes, A. R., Nickolson, L., & Doosje, B. (2015). Triggerfactoren in het radicaliseringsproces.
Amsterdam: Universiteit van Amsterdam.
Feldman, S., & Stenner, K. (1997). Perceived threat and authoritarianism. Political Psychology,
18(4), 741–770.
Fiske, S. T., Cuddy, A. J., Glick, P., & Xu, J. (2002). A model of (often mixed) stereotype content:
Competence and warmth respectively follow from perceived status and competition. Journal of
Personality and Social Psychology, 82(6), 878–902.
Gallagher, M. J. (2014). Terrorism and organised crime: Co-operative endeavours in Scotland?
Terrorism and Political Violence, 26(2), 320–336.
Glick, P. (2002). Sacrificial lambs dressed in wolves’ clothing. In L. S. Newman & R. Erber (Eds.),
Understanding genocide: The social psychology of the Holocaust (pp. 113–142). London:
Oxford University Press.
170
V. Ljujic et al.
Huddy, L., Feldman, S., & Weber, C. (2007). The political consequences of perceived threat and
felt insecurity. The Annals of the American Academy of Political and Social Science, 614(1),
131–153.
Hutchinson, S., & O’Malley, P. (2007). A crime–terror nexus? Thinking on some of the links
between terrorism and criminality. Studies in Conflict Terrorism, 30(12), 1095–1107.
Joosen, K., & Slotboom, A. M. (2015). Terugblikken op de aanloop: Dynamische voorspellers van
perioden van detentie gedurende de levensloop van vrouwelijke gedetineerden in Nederland.
Tijdschrift voor Criminologie, 57(1), 84–98.
Kakar, S. (1996). The colors of violence: Cultural identities, religion, and conflict. Chicago:
University of Chicago Press.
King, M., & Taylor, D. M. (2011). The radicalization of homegrown jihadists: A review of theoretical models and social psychological evidence. Terrorism and Political Violence, 23(4),
602–622.
Klausen, J. (2015). Tweeting the Jihad: Social media networks of Western foreign fighters in Syria
and Iraq. Studies in Conflict & Terrorism, 38(1), 1–22.
Koomen, W., & van der Pligt, J. (2009). Achtergronden en determinanten van radicalisering en
terrorisme. Tijdschrift voor Criminologie, 51(4), 345–359.
Kruglanski, A. W., Gelfand, M. J., Bélanger, J. J., Sheveland, A., Hetiarachchi, M., & Gunaratna,
R. (2014). The psychology of radicalization and deradicalization: How significance quest
impacts violent extremism. Political Psychology, 35(S1), 69–93.
Kruglanski, A. W., Jasko, K., Chernikova, M., Dugas, M., & Webber, D. (2017). To the fringe
and back: Violent extremism and the psychology of deviance. American Psychologist, 72(3),
217–230.
Li, Q., & Brewer, M. B. (2004). What does it mean to be an American? Patriotism, nationalism,
and American identity after 9/11. Political Psychology, 25(5), 727–739.
Ljujic, V., van Prooijen, J. W., & Weerman, F. (2017). Beyond the crime-terror nexus: Socio-­
economic status, violent crimes and terrorism. Journal of Criminological Research, Policy and
Practice, 3(3), 158–172.
Makarenko, T. (2012). Foundations and evolution of the crime–terror nexus. In F. Allum & S.
Gilmour (Eds.), Routledge handbook of transnational organized crime (pp. 252–267). New
York: Routledge.
McCauley, C., & Moskalenko, S. (2008). Mechanisms of political radicalization: Pathways toward
terrorism. Terrorism and Political Violence, 20(3), 415–433.
Moghaddam, F. (2005). The staircase to terrorism: A psychological exploration. American
Psychologist, 60(2), 161–169.
Neuberg, S. L., & Cottrell, C. A. (2006). Evolutionary bases of prejudices. In M. Schaller, J. A.
Simpson, & D. T. Kenrick (Eds.), Evolution and Social Psychology (pp. 163–187). New York:
Psychology Press.
Neuberg, S. L., Kenrick, D. T., & Schaller, M. (2011). Human threat management systems: Self-­
protection and disease avoidance. Neuroscience & Biobehavioral Reviews, 35(4), 1042–1051.
Pare, P. P., & Felson, R. (2014). Income inequality, poverty and crime across nations. The British
Journal of Sociology, 65(3), 434–458.
Ranstorp, M. (1996). Terrorism in the name of religion. Journal of International Affairs, 50(1),
41–62.
Riek, B. M., Mania, E. W., & Gaertner, S. L. (2006). Intergroup threat and outgroup attitudes: A
meta-analytic review. Personality and Social Psychology Review, 10(4), 336–353.
Sageman, M. (2004). Understanding terror networks. Philadelphia: University of Pennsylvania
Press.
Sageman, M. (2008). The next generation of terror. Foreign Policy, 165(March/April), pp. 36–42.
Schmid, A. P. (2004). Frameworks for conceptualising terrorism. Terrorism and Political Violence,
16(2), 197–221.
Schmid, A. P. (2011). The Routledge handbook of terrorism research. London/New York:
Routledge.
Testing a Threat Model of Terrorism: A Multi-method Study About Socio-Economic…
171
Schmid, K., & Muldoon, O. T. (2015). Perceived threat, social identification, and psychological
well-being: The effects of political conflict exposure. Political Psychology, 36(1), 75–92.
Schmitt, D. P., & Pilcher, J. J. (2004). Evaluating evidence of psychological adaptation: How do
we know one when we see one? Psychological Science, 15(10), 643–649.
Shadid, W. A. (1991). The integration of Muslim minorities in the Netherlands. International
Migration Review, 25, 355–374.
Staub, E. (2004). Understanding and responding to group violence: Genocide, mass killing, and terrorism. In F. M. Moghaddam & A. J. Marsella (Eds.), Understanding terrorism: Psychosocial
roots, consequences, and interventions (pp. 151–168). Washington: American Psychological
Association.
Stephan, W. G., & Stephan, C. W. (2005). In J.F. Dovidio, Glick, P., & Rudman, L.A. (Eds.)
Intergroup relations program evaluation. On the nature of prejudice: Fifty years after Allport
(pp. 431–446). Malden, MA: Blackwell Publishing.
Stephan, C. W., & Stephan, W. S. (2013). An integrated threat theory of prejudice. In S. Oskamp
(Ed.), Reducing prejudice and discrimination (pp. 33–56). New York/London: Psychology
Press.
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. Austin & S.
Worchel (Eds.), The social psychology of intergroup relations (pp. 33–47). Monterey: Brooks/
Cole.
Van Ham, T., Hardeman, M., Van Esseveldt, J., Lenders, A., & Van Wijk, A. (2018). A view on
left-wing extremism: An exploratory study into left-wing extremist groups in the Netherlands.
Den Haag: WODC.
van Prooijen, J. W., Krouwel, A. P., & Pollet, T. V. (2015). Political extremism predicts belief in
conspiracy theories. Social Psychological and Personality Science, 6(5), 570–578.
Veldhuis, T., & Bakker, E. (2009). Muslims in the Netherlands: Tensions and violent conflict
(MICROCON Policy Working Paper 6). Brighton: MICROCON.
Vellenga, S. (2018). Anti-Semitism and Islamophobia in the Netherlands: Concepts, developments, and backdrops. Journal of Contemporary Religion, 33(2), 175–192.
Verkuyten, M. (2018). Religious fundamentalism and radicalization among Muslim minority
youth in Europe. European Psychologist, 23(1), 21–31.
Versteegt, I., Ljujic, V., El Bouk, F., Weerman, F., & van Maanen, F. (2018). Terrorism, adversity
and identity. A qualitative study on detained terrorism suspects in comparison to other detainees. Amsterdam: NSCR.
Victoroff, J. E., & Kruglanski, A. W. (2009). Psychology of terrorism: Classic and contemporary
insights. New York: Psychology Press.
Visser-Vogel, E., de Kock, J., Bakker, C., & Barnard, M. (2018). I have Dutch nationality, but
others do not see me as a Dutchman, of course. Journal of Muslims in Europe, 7(1), 94–120.
Weenink, A. W. (2015). Behavioral problems and disorders among radicals in police files.
Perspectives on Terrorism, 9(2), 17–33.
Weggemans, D., Bakker, E., & Grol, P. (2014). Who are they and why do they go? The radicalization and preparatory processes of Dutch jihadist foreign fighters. Perspectives on Terrorism,
8(4), 100–110.
Wittendorp, S., de Bont, R., Bakker, E., & de Roy van Zuijdewijn, J. (2017). Persoonsgerichte
aanpak Syrië/Irak-gangers. Leiden: Universiteit Leiden.
Zick, A., Pettigrew, T. F., & Wagner, U. (2008). Ethnic prejudice and discrimination in Europe.
Journal of Social Issues, 64(2), 233–251.
Part II
Factors Affecting Recruitment to
Organized Crime
Systematic Review of the Social,
Psychological and Economic Factors
Relating to Involvement and Recruitment
into Organized Crime
Tommaso Comunale, Francesco Calderoni, Martina Marchesi,
Elisa Superchi, and Gian Maria Campedelli
Background: Recruitment into Organized Crime
Organized crime (henceforth OC) differs from general and serious crime due to the
higher complexity of the criminal activities, often requiring a network of transnational contacts to identify suitable and trustworthy co-offenders or contacts with the
legal world (Cornish and Clarke 2002; Kleemans and De Poot 2008). The relevance
of social ties among OC offenders and the social environment constitutes the heart
of early studies on criminal organizations as the mafias (Albini 1971; Cressey 1969;
Ianni 1974). The social embeddedness of OC drove more recent analyses on the
sociocultural processes shaping recruitment pathways and criminal careers of OC
offenders (Kleemans and De Poot 2008; Kleemans and Van de Bunt 1999, 2008;
Kleemans and Van Koppen 2014). The complex dynamics underlying individuals’
involvement into OC have contributed to the resilience of organized crime groups
(henceforth OCGs), one of their most distinct features. The ability of rapidly reorganizing and constantly recruiting new members has also favored the consolidation
of OCGs in some geographic areas (e.g. the mafias in Italy). Literature on recruitment into OC has not only focused on social factors for involvement into OCGs, but
scientific enquiries have also investigated psychological (Ostrosky et al. 2012) and
economic factors (Lavezzi 2008, 2014).
As for social factors, people get involved into OC through social and work ties.1
Social ties with co-offenders and with the legal world play a central role in the success
of OC-related activities. Furthermore, social ties may create a “social snowball effect”,
Involvement into OC also takes place through leisure activities, because of life events or through
deliberate recruitment by criminal organizations. For more details, see Kleemans and Van Koppen
(2014).
1
T. Comunale (*) · F. Calderoni · M. Marchesi · E. Superchi · G. M. Campedelli
Università Cattolica del Sacro Cuore and Transcrime, Milan, Italy
e-mail: tommaso.comunale@unicatt.it
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_8
175
176
T. Comunale et al.
a process in which people get involved in OC through people that are close to them
(Kleemans and Van Koppen 2014, p. 288). Relations between individuals bridge
social and criminal networks – also across countries – providing new criminal opportunities and solve problems of cooperation in hostile environments in which offenders
carry out illegal activities. Opportunities for committing OC may also arise through
work ties. Occupation settings offer good opportunities to gain trust in someone else,
essential for activities involving financial and other risks. For instance, working in the
mobility, transport, and logistics sector may provide opportunities for cross-border
OC activities. Criminal co-operations therefore may be established through peoples’
international contacts and travel movements; through their individual freedom of
movement; and through their occupations enabling them to meet with potential cooffenders (Kleemans and Van de Bunt 2008). Once part of a criminal network, one’s
criminal career can take different paths depending on his/her roles and responsibilities. In this regard, Kleemans and de Poot (2008) highlight four major roles: offenders
with strong local roots, individuals specialized in some activities, individuals obtaining increasing roles through capital accumulation, and offenders who possess specific
skills or expertise that make the other offenders dependent from them. For these reasons, recruitment into OCGs and criminal careers can also take place during adulthood among late starters – i.e. adult members with no previous criminal involvement
(Kleemans and Van Koppen 2014). Thus, a career in OC does not necessarily start in
early adolescence, where individual characteristics and long-term risk factors may
serve as an explanation for a lifetime involvement in crime.
Psychological factors have also been assumed to influence individuals’ recruitment
into OCGs. Information on psychological traits of OC offenders, however, is limited
because of the relatively small population, which is difficult to reach also in contexts
of detention. Prior studies about OC relied on the Psychopathy Check List-­Revised
(Ostrosky et al. 2012), a controversial instrument resting on the contested concept of
psychopathy, bearing a number of limitations (Jones et al. 2013). Other neuroscientific tools such as neuropsychological and genetic tests, as well as neuroimaging, are
widely used in courts to explain and sometimes justify criminal behavior (Walsh and
Beaver 2009). To advance the understanding of the psychological determinants of
criminal behavior, such tools might be adopted for criminological purposes. Moreover,
when investigating the psychological factors for the recruitment into OCGs, researchers may also take into consideration other types of disorders, e.g. substance abuse
disorders, low self-control and/or a history of past disorders and negative development.
Economic risk factors have also emerged as relevant in addressing the processes
shaping individuals’ recruitment into OCGs. Lavezzi (2008, 2014), for instance,
identified several aspects that make Sicily’s economy vulnerable to OC penetration.
Extortion and protection are common in Sicily, whose economy is characterized by
a large proportion of small firms, large sectors of traditional/low-tech economic
activity, a large construction sector, and a large public sector (Lavezzi 2008).
Opportunities for OCGs also emerge from impairments to the rule of law and the
presence of illegal and informal markets (Bandiera 2003). Fluctuations and
­irregularities in the credit markets, for example, may create opportunities for usury.
In addition, inequality has been indicated as a possible facilitator in the emergence
of OCGs. Data from Italian regions show a remarkable correlation between high-­
Systematic Review of the Social, Psychological and Economic Factors Relating…
177
income inequality and the spread of OC (Daniele 2009). The majority of scientific
studies has focused on the relation between economic risk factors and OC at the
macro-level, but economic disadvantages can also influence the individuals’ propensity to join OCGs (Carvalho and Soares 2016).
Despite the increasing number of studies on several aspects of OC, to date little
systematized information is available on the social, psychological and economic
factors that lead to involvement in OC. This systematic review embraces a broad
perspective encompassing several aspects related to recruitment into OCGs (e.g.
age, gender, social ties, economic conditions, criminal background and skills).2
In line with these purposes, this systematic review firstly aims at identifying the
current knowledge on social, psychological, and economic factors relating to
recruitment to OCGs. It will also highlight whether these factors are independent of
one other or they are correlated. Secondly, this review aims at assessing the validity
and generalizability of research findings of studies employing different research
methods and focusing on different countries. In this regard, this systematic review
seeks to answer two sets of research questions:
• What are the most commonly reported social, psychological, and economic factors leading to criminalisation and recruitment into OC networks? Are them to
be viewed independently of one another?
• Which methods do included studies apply? Are finding consistent through studies
employing different methods and adopting different geographic scopes?
The systematic assessment of the empirical evidence on such factors contributes
to the consolidation of the knowledge on involvement into OC and may inform the
policy making process. In addition, research findings will serve to set direction for
future enquiries aimed at revealing the underlying dynamics of recruitment
into OCGs.
Methodology
Operational Definitions of Organized Crime
This study aims at revealing the factors associated with recruitment into OC. The
concept of recruitment refers to the different processes leading individuals to the
stable involvement into OCGs. While this interpretation is broader than the formal
affiliation to any OCG, it excludes individuals occasionally cooperating or co-­
offending with OCGs.
Previous systematic reviews have focused on youth gang membership and interventions
(Hodgkinson et al. 2009; Klein and Maxson 2006). The Campbell Collaboration, for example, has
published two systematic reviews on the involvement of young people in gangs (Fisher et al.
2008a, b), and more recently one on predictors of youth gang membership in low- and middleincome countries (Higginson et al. 2014). These reviews did not consider the factors relating to
membership in other types of groups involved in crime, namely OCGs.
2
178
T. Comunale et al.
As for organized crime, divergent approaches adopted by governments and policymakers for long resulted in the difficulty to produce a uniform definition (see
Paoli 2014a).
This systematic review adopts the definition of organized crime (OC) by the
United Nations Convention against Transnational Organized Crime (United Nations
2000, p. 5):
‘Organized criminal’ group shall mean a structured group of three or more persons, existing for a period of time and acting in concert with the aim of committing one or more serious crimes or offences established in accordance with this Convention, in order to obtain,
directly or indirectly, a financial or other material benefit.
The UN definition is generally accepted from a legal perspective, though some
scholars have criticized it for being rather generic and unspecific (Albrecht and
Fijnaut 2002; Paoli 2014a). Despite limitations, the UN definition encompasses different forms of OC, allowing this systematic review to include several types of
OCGs: mafias, DTOs, gangs, and a residual category labelled “other criminal
organizations”.
Mafias are characterized by a remarkable longevity, an organizational and cultural complexity enforcing the respect of the code of silence (i.e. omertà). In addition, mafias’ ability to control legitimate markets and to exercise a political dominion
over their areas of settlement, frequently using private violence and simultaneously
supplying protection against extortion, competitors and law enforcement
(Finckenauer 2007; Gambetta 1993; Paoli 2014b; Varese 2010). DTOs are complex
organizations involved in the production, transport and/or distribution of large
quantities of illicit drugs through a business-like structure (Clapper 2015; Knox and
Gray 2014; U.S. Department of Justice 2010). As regards to gangs, given the important share of adult offenders in OC and the relevance of the ties to the legitimate
world, this study excludes youth (street) gangs and prison gangs in favor of criminal
gangs composed of adults with different degrees of organizational structure.3 This
study therefore considers “criminal gangs” those groups of adults following a code
of conduct, having common beliefs and existing in a semi- structured organization
oriented to accomplish their goals through criminal activity (Langston 2003). Lastly,
the label “other criminal organizations” encompasses different groups that “utilize
violence or threats of violence, provide illicit goods that are in public demand, and
assure immunity for their operators through corruption and enforcement” (Hagan
2015, p. 395). These comprise for instance outlaw motorcycle gangs and other
OCGs active in specific world regions (e.g. the Balkans).
This choice is also due to the fact that the literature generally considers youth street gangs as different from OCGs (Decker and Pyrooz 2014). Furthermore, a recent systematic review has already
assessed the factors leading to youth gang membership (Higginson et al. 2014). Similarly, while
there is a relevant literature on prison gangs, this field is mostly separate from the literature on OC,
which emphasizes the social embeddedness and the role of ties with the legitimate world. Prison
gangs, on the contrary, occur in a specific and institutionalized settings and recruitment is influenced by contextual factors (Blevins et al. 2010; Wood et al. 2014).
3
Systematic Review of the Social, Psychological and Economic Factors Relating…
179
Search Strategy and Selection of the Relevant Studies
This systematic review relies on academic and grey literature retrieved from 12
electronic databases in English, French, German, Italian, and Spanish.4 The databases include different research fields, without limitations as to their year of publication or geographic origin.
A three-fold query structure incorporated information on social, psychological,
and economic factors relating to the recruitment into mafias, DTOs, gangs, and
other criminal organizations. The queries combined search terms from each of the
three main categories, i.e. type of OCGs, the type of factors, and recruitment. The
Boolean operator “OR” connected keywords of the same category, while the
Boolean operator “AND” connected keywords from different categories. This query
structure ensured the collection of all the studies indexed in the electronic databases
containing at least one term from each semantic category.
The search led to the collection of 48,731 unique studies. A systematic screening
of these studies selected the relevant information according to shared eligibility criteria. Firstly, a team of trained researchers screened the titles and abstracts of each
study to include only the literature making an original research contribution (e.g.
excluding news articles or reviews of any type) to the social, psychological, and economic factors relating to the recruitment into OCGs. Of the initial studies, 118 met
the inclusion criteria. This initial list of studies was then integrated with additional
literature from their bibliographies, as well as with the suggestions of several experts
in the field of OC.5 These activities led to a total of 130 studies ­available for full-text
screening. Secondly, the full-text screening further selected the studies according to
stricter criteria. Only empirical studies were included. Of the remaining 57 empirical
studies, 47 studies had a clear research aim and an appropriate research methodology, design, recruitment strategy, and data collection.6
This systematic review included studies adopting quantitative, qualitative, or
mixed-methods approaches. Systematic reviews have generally excluded qualitative
research although there has been much debate on its incorporation (see Dixon-­
English: Criminal Justice Abstracts; Open Grey; Social Sciences Premium, NJCRS, PsycInfo,
Abi/Inform, International Bibliography of the Social Sciences, Public Health Database, Military
Database, EconLit, PsycArticles; PubMed; Scopus; Web of Science. French: Google Scholar;
Sudoc.Abes. German: Sowiport. Italian: Riviste Web. Spanish: Liliacs; Latin America & Iberia
Database.
Dutch was excluded from the search as contacts with Dutch scholars confirmed that empirical
studies in this field have also been published in English.
5
The experts that contributed to this systematic review are: Jay Albanese (Virginia Commonwealth
University, USA), Paolo Campana (University of Cambridge, UK), Scott Decker (Arizona State
University, USA), Edward Kleemans (Vrije University of Amsterdam, NL), Klaus Von Lampe
(John Jay College of Criminal Justice, USA), Carlo Morselli (University of Montreal, CA), Letizia
Paoli (Katholieke Universiteit Leuven, BE), David Pyrooz (University of Colorado Boulder,
USA), Sonja Wolf (Centro de Investigación y Docencia Económicas, MEX).
6
The included studies meet the quality criteria set out by an adapted version of the Critical
Appraisal Skills Programme (2017) for qualitative studies.
4
180
T. Comunale et al.
Woods et al. 2006). The reason relies on the fact that systematic reviews aim at
studying correlates and risk factors–or at addressing the effectiveness of intervention programs–which require quantitative studies as the primary basis (for risk factors, see Murray et al. 2009). Excluding qualitative contributions from the review,
however, could have impacted the research findings, as an important amount of
organized crime studies employ either qualitative or mixed-methods approaches.
This systematic review therefore considered the inclusion of such methods to provide a more comprehensive, albeit descriptive, overview about the factors leading to
recruitment into OCGs.7
In addition, this study relied on grey literature to validate the research findings.
Inclusion of grey literature allows to “take into account some important contextual
information, without losing the level of rigor required for a systematic review”
(Benzies et al. 2006, p. 59). The grey literature considered all available reports
issued by the Italian governmental, judicial, and law enforcement institutions on the
presence of the Italian mafias in Italy and abroad, namely the Italian Parliamentary
Anti-Mafia Commission (CPA), the Italian National Anti-Mafia and Counterterrorism
Directorate (DNA), and Anti-Mafia Investigative Directorate (DIA). These sources
are relevant as the mafias have some peculiarities differentiating them from the
other types of OCGs. Only reports containing relevant information for the purposes
of this review were used to corroborate results, namely 8 reports issued by the CPA,
9 by the DNA, and 23 by the DIA.
Results
Description of the Included Studies
The included studies were published between 1969 and 2017, spanning different
geographical regions and languages. They also differ according to the type of
sources: 6% (n = 3) are dissertations, 36% (n = 17) are books or chapters, and 58%
(n = 27) are journal articles. The articles come from 19 journals, mostly pertaining
to criminology and related fields of studies.8 The included studies also differ regard-
This systematic review was conducted following in part the guidelines provided by the Campbell
Collaboration–an international research network promoting and producing systematic reviews–as
Campbell reviews are based solely on quantitative findings as primary basis for conclusions. For
more details, see https://www.campbellcollaboration.org/library/campbell-collaboration-systematic-reviews-policies-and-guidelines.html
8
The journals are: British Journal of Criminology; Crime Law & Social Change; Trends in
Organized Crime; European Journal of Criminology; Global Crime; Canadian Journal of
Criminology; Criminal Behaviour and Mental Health; Criminology; European Sociological
Review; International Journal of Social Psychology; Journal of Criminal Justice; Journal of
Economic Behavior & Organization; Justice Policy Journal; Law & Society Review; Policing: An
International Journal of Police Strategies & Management; Research and Reports in Forensic
Medical Science; Revista Electrónica de Ciencia Penal y Criminología; Social Problems; The
Howard Journal of Crime and Justice.
7
Systematic Review of the Social, Psychological and Economic Factors Relating…
181
ing the type of OCGs they focus on. Almost half (42%, n = 20) deal with the mafias,
especially with the Italian ones, 30% (n = 14) discuss about other criminal organization, above all Dutch ones, and the remaining about gangs (15%, n = 7), and DTOs
(13%, n = 6).
The selected studies rely on different types of data collection and methods of
analysis. In investigating the social, psychological, and economic factors relating to
the recruitment into OCGs, they use both primary and secondary data, and often the
combination of the two (Fig. 1, on the left). Due to the difficulties to study the
recruitment into OCGs, almost half of the studies (49%, n = 23) relied on the information coming from official data, investigative and judicial files, as well as other
documents (e.g. bibliographies and tapes). The remaining half of the studies either
integrated information from secondary data with interviews (23%, n = 11), or conducted their research entirely through surveys, interviews, and conversations with
informants (28%, n = 13).
The studies applied different methods of analysis (Fig. 1, on the right). More
than half of them (57%, n = 27) favored a qualitative approach in dealing with the
topic of factors relating to the recruitment into OCGs. The remaining adopted either
quantitative (26%, n = 12) or mixed-methods approaches (17%, n = 8) (see Tables
1, 2, and 3, respectively). The scarcity of quantitative studies may be due to the topic
of the research itself, i.e. the recruitment into OCGs, is not particularly suitable for
such methodological approach. The quantitative approach is further hindered by the
difficulties in operationalizing and measuring the social, psychological, and economic factors intervening in the recruiting process.
28%
23%
49%
Primary data
Secondary data
Primary and secondary data
17%
26%
57%
Quantitative
Qualitative
Mixed-methods
Source: authors’ elaboration of the studies included in this systematic review.
Fig. 1 Types of data collection and methods of analysis (n = 47). (Source: authors’ elaboration of
the studies included in this systematic review.)
190
4109
DTOs
Jhi and Gerber Gangs
(2015)
Other Criminal
Organizations
Gangs
Gangs
Carvalho and
Soares (2016)
Kirby et al.
(2016)
Kissner and
Pyrooz (2009)
May (2009)
138
200
230
Type of OCG
Other Criminal
Organizations
Source
Blokland et al.
(2017)
Survey
Interviews
Official data
Survey
Interviews
Data
Samplea collection
601
Official data
Table 1 Details on the included quantitative studies
Descriptive statistics
Correlation
Logistic regression
Logistic regression
Descriptive statistics
Chi square
Kruskal-­Wallis test
Descriptive statistics
Logistic regression
Data analysis
Descriptive statistics
T-test
Chi square
Logistic regression
Descriptive statistics
Mincerian regression
Categories of
factors
Age
Criminal
background and
skills
Age
Ethnicity
Educational
background
Economic
conditions
Employment
Economic
conditions
Age
Gender
Ethnicity
Criminal
background and
skills
Age
Social ties
Psychological
factors
Group identity
Psychological
factors
Gang membership is positively
associated with unemployment
Compared with general crime
offenders, OC offenders are more
predominantly male, more
ethnically heterogeneous and
with more drug offences records
Persistent gang involvement is
associated with poor self-control
Gangs constitute a form of social
support for members
U.K.
U.S.
U.S.
Relevant findings
Outlaw motorcycle gangs’
membership is positively
associated with previous criminal
records
Individuals join DTOs for
monetary returns
U.S.
Brazil
Geographic
scope
Netherlands
182
T. Comunale et al.
2384
69
230
222
OCGs
Mafias
Salinas and
Regadera
(2016)
Schimmenti
et al. (2014)
Unlu and Ekici DTOs
(2012)
Wang (2013)
Interviews
Investigative
files
Interviews
Investigative
files
Investigative
files
Interviews;
Documents
Content analysis
CHAID
Logistic regression
Descriptive statistics
T-test
Chi square
Logistic regression
Descriptive statistics
Chi square
Descriptive statistics
Descriptive statistics
Descriptive statistics
a
The sample reports the number of individuals included in the studies
DTOs
200
OCGs
Requena et al.
(2014)
82
OCGs & DTOs
Ostrosky et al.
(2012)
Age
Gender
Economic
conditions
Ethnicity
Employment
Economic
conditions
Social ties
Age
Economic
conditions
Psychological
factors
Gender
Social ties
Silence/Omertà
Age
Gender
Ethnicity
Social ties
Criminal
background and
skills
Psychological
factors
High levels of antisocial traits
and low levels of interpersonalaffective traits of psychopathy
characterize mafia members
Male adults from low-income
societies can join DTOs and
work as couriers for monetary
returns
Early starters get involved in
DTOs because of entertainment
expenses, while late starters
because of financial difficulties
Italy
U.K.
Turkey
Spain
Spain
Individuals join DTOs seeking
for higher economic income and
a lifestyle characterized by the
accumulation/showing off of
material values
Women get involved in OCGs
through their personal networks,
i.e. family ties and emotional ties
Individuals get involved in OCGs
because they possess special
expertise developed outside the
criminal world
Mexico
Systematic Review of the Social, Psychological and Economic Factors Relating…
183
Interviews
Investigative files
Investigative files
Judicial records
Interviews
Documents
Mafias
Behan (1996)
Brancaccio (2017) Mafias
Brotherton and
Barrios (2004)
Gangs
Other Criminal Interviews
Organizations Investigative files
Judicial records
Arsovska (2015)
Interviews
Investigative files
Judicial records
Mafias
Arlacchi (1983)
Data collection
Interviews
informants
Documents
Type of OCG
Mafias
Source
Albini (1971)
Table 2 Details on the included qualitative studies
Categories of factors
Educational background
Economic conditions
Social ties
Group identity
Criminal background and skills
Silence/Omertà
Age
Educational background
Social ties
Group identity
Criminal background and skills
Age
Ethnicity
Economic conditions
Social ties
Group identity
Criminal background and skills
Age
Economic conditions
Social ties
Criminal background and skills
Employment
Economic conditions
Social ties
Criminal background and skills
Economic conditions
Group identity
Individuals are recruited into OCGs based on
ethnic ties, kinship ties, and display of violent
behavior
Incarcerated individuals and especially
youngsters attracted by the cult of violence
can join the mafias
Individuals enter the mafias because of
kinship and blood ties, coupled with low
economic conditions and violent/risk taking
behavior
While females join gangs through blood ties,
males rely on kinship and social relations
developed in prison
Balkans
Italy
Italy
U.S.
Individuals who get involved in mafias come
from the middle-class, have an educational
background and possess managerial skills
Italy
Geographic scope Main findings
Italy
Recruitment of individuals into OCGs is also
based on friendship, kinship, contract, and
patron-client relationships
184
T. Comunale et al.
Type of OCG
Mafias
Mafias
DTOs
Gangs
Mafias
DTOs
Gangs
Source
Ciconte (1992)
Cressey (1969)
Decker and
Chapman (2008)
Densley (2012)
Gambetta (1993)
García (2006)
Gordon (2000)
Interviews
Tapes and CDs
Interviews
Judicial records
Ethnicity
Economic conditions
Social ties
Canada
(continued)
Individuals get involved in DTOs because of
their need of power, belonging, respect,
security, and pride
Individuals often access gangs through close
friends due to ethnic marginality and the
attraction of supportive peer groups
Individuals are recruited into gangs based on
personal features like criminal competency
and group loyalty
Being able to enforce the code of silence and
possessing relevant expertise are crucial
factors for being recruited into the mafias
Ethnic ties have a crucial role in DTOs
membership
U.S.
Latin America
U.K.
Values like honor, loyalty, and silence are
crucial for individuals aiming at joining the
mafias
U.S.
Geographic scope Main findings
Italy
Kinship and blood ties and the lack of
legitimate occupations facilitate the
involvement into the mafias
Italy
Employment
Social ties
Group identity
Criminal background and skills
Silence/Omertà
Group identity
Mexico
Categories of factors
Age
Ethnicity
Educational background
Employment
Economic conditions
Social ties
Criminal background and skills
Silence/Omertà
Age
Interviews
Investigative files Ethnicity
Judicial records Social ties
Group identity
Criminal background and skills
Silence/Omertà
Interviews
Ethnicity
Economic conditions
Social ties
Criminal background and skills
Interviews
Group identity
Criminal background and skills
Data collection
Judicial records
Documents
Systematic Review of the Social, Psychological and Economic Factors Relating…
185
Type of OCG
Mafias
Gangs
Mafias
Mafias
Mafias
Mafias
Mafias
Source
Hess (1993)
Hixon (2010)
Lo (2010)
Lupo (1993)
Paoli (2003)
Sales (2015)
Sciarrone (2014)
Table 2 (continued)
Judicial records
Judicial records
Documents
Interviews
Documents
Judicial records
Documents
Documents
Interviews
Data collection
Judicial records
Documents
Italy
Educational background
Employment
Economic conditions
Social ties
Criminal background and skills
Social ties
Italy
China
Social ties
Group identity
Criminal background and skills
Silence/Omertà
Italy
Ethnicity
Social ties
Criminal background and skills
Silence/Omertà
Italy
Social ties
Group identity
Silence/Omertà
Categories of factors
Geographic scope
Italy
Educational background
Economic conditions
Social ties
Group identity
Criminal background and skills
Silence/Omertà
U.S.
Gender
Group identity
Psychological factors
The mafias mainly recruit new members
based blood and kinship ties which are
reinforced through an initiation ceremony
with specific rituals
The contract of fraternization among
recruited mafia members extends the bonds
of loyalty and obligation beyond family ties,
providing mutual support and trust
Individuals capable of strategic use of
violence are recruited into the mafias
The mafias mainly recruit new members
based on ethnicity and kinship and blood ties
Individuals recruited into OCGs usually have
an history of negative and arrested
development resulting in antisocial
personality disorders
Being able to connect people from different
environments favor recruitment into Chinese
Triads
Main findings
Individuals get involved in the mafias through
social relations, especially because of their
ability to connect the criminal members with
other third parties, like other criminal groups
or representative of legal businesses
186
T. Comunale et al.
Zhang and Chin
(2002)
Varese (2011b)
Varese (2006)
Varese (2001)
Van San and
Sikkens (2017)
Van Koppen
(2013)
Source
Sergi (2016)
Geographic scope Main findings
Italy
Being part of an OCG results from cultural
transmission and gradual learning, therefore
kinship and blood ties have a crucial role in
the recruitment of new members
Netherlands
Individuals engage in OC activities exploiting
their skills and kinship and blood ties, as
criminals prefer to unfold their activities with
family members or close friends
Netherlands
Female smugglers join DTOs mainly through
DTOs
Interviews
Gender
Peru
their personal networks, i.e. family ties,
Informants
Economic conditions
romantic relationships, and friendships
Social ties
Social ties
Russia
Individuals getting involved in the mafias are
Mafias
Interviews
recruited from a pool of trusted aspirants with
Investigative files Criminal background and skills
no previous connections with lawJudicial records
enforcement agents
Documents
Mafias
Documents
Social ties
Italy
The mafias kin-based system of recruitment
facilitates transplantation of criminal
activities in a new region: when an entire
blood family migrates the criminal group
automatically reconstitutes itself
When Russian mafia groups migrates, they
Italy
Ethnicity
Mafias
Interviews
recruit new local members based on their
Hungary
Judicial records Social ties
dependability and proven ability to use
Criminal background and skills U.S.
Documents
violence
China
Other Criminal Interviews
Ethnicity
U.S.
Individuals enter Chinese Triads because of
Organizations
Group identity
China
their commitment to making money and
direct connections in the Chinese
communities
Type of OCG
Data collection
Categories of factors
Other Criminal Investigative files Economic conditions
Organizations Documents
Social ties
Group identity
Psychological factors
Other Criminal Investigative files Employment
Organizations
Economic conditions
Social ties
Systematic Review of the Social, Psychological and Economic Factors Relating…
187
Other Criminal Investigative files
Organizations
Mafias
Other Criminal Interviews
Organizations Survey
Other Criminal Investigative files
Organizations
Other Criminal Investigative files
Organizations
Mafias
Mafias
Kleemans and Van
de Bunt (2008)
Morselli (2003)
Salinas, Requena
and de la Corte
(2011)
Van Koppen et al.
(2010b)
Van Koppen et al.
(2010a)
Varese (2011a)
Varese (2013)
Investigative files
Investigative files
Investigative files
Documents
Type of OCG
Data collection
Other Criminal Investigative files
Organizations
Source
Kleemans and de
Poot (2008)
Table 3 Details on the included mixed-methods studies
Gender
Ethnicity
Social ties
Gender
Ethnicity
Social ties
Age
Gender
Employment
Social ties
Criminal background and skills
Age
Criminal background and skills
Criminal background and skills
Social ties
Criminal background and skills
Categories of factors
Age
Employment
Social ties
Criminal background and skills
Employment
Social ties
Netherlands Most of individuals get involved in OCGs when
they are adults
Netherlands Compared with general crime offenders, OC
offenders more often have previous and serious
judicial records
Italy
Women getting involved in the mafias have a
more relevant role abroad than in their territory
of origin
Italy
Individuals getting involved in the mafias
abroad mainly focus on economic investments
and resource acquisition
Netherlands Social relations in leisure and work settings
may provide opportunities for joining OCGs
throughout individuals’ lives
U.S.
Social relations and their management are
crucial in the involvement and career in the
mafias
Spain
Individuals join OCGs because their
employment could favor illegal activities or to
have an additional income to their salary
Geographic
scope
Main findings
Netherlands Social relations in leisure and work settings
may provide opportunities for joining OCGs
throughout individuals’ lives
188
T. Comunale et al.
Systematic Review of the Social, Psychological and Economic Factors Relating…
189
Table 4 Categories of socio-economic and psychological factors associated with recruitment into
OCGs and number of studies reporting on factor categories (n = 184)
Category of factors
Age
Gender
Ethnicity
Educational
background
Employment
Economic conditions
Social ties
Group identity
Psychological factors
Criminal background
and skills
Silence/Omertà
Explanation of included factors
Factors on the age at which individuals are likely to be
recruited into OCGs (e.g. youngsters)
Factors on the differences between genders in being involved
in OCGs (e.g. majority of OCGs members are males)
Factors on the ethnic features of OCGs members (e.g. ethnic
homogeneity)
Factors on the level of education of individuals recruited by
OCGs (e.g. illiteracy)
Factors on the how the lack of employment or the presence
of specific work settings can be facilitators for joining OCGs
(e.g. employees of logistic companies)
Factors on how certain economic conditions can facilitate
recruitment into OCGs (e.g. poverty)
Factors on the relations that provide opportunities for being
involved in OCGs (e.g. family ties)
Factors on the values and subculture shared by OCGs
members
(e.g. honor)
Factors on disorders related to the recruitment into OCGs
(e.g. being antisocial)
Factors on how previous criminal records, specific skills and
attitudes can increase the opportunities to enter OCGs
(e.g. risk-taking behaviors)
Factors on the ability of enforcing the code of silence for
being part of OCGs
N. of
factors
17
12
13
7
20
19
38
18
6
35
9
Notwithstanding the differences among the included studies, each of them
reported on the social, psychological, and/or economic factors associated with
recruitment of individuals to OCGs. After carefully reading all the studies, the
research team created categories of factors to organize the relevant information and
systematize the knowledge on this field of study. The total number of factors identified across the 47 included studies is 184. The identification of factors in each study
and their systematization into 11 categories allows analyzing how many times they
are reported. The recurring factors are those related to social ties and criminal background and skills, which respectively appear 38 and 35 times throughout the
included studies (Table 4).
Figure 2 highlights the different methodological approaches adopted by the studies reporting on each category of factors. As already underlined, the qualitative
approach seems to fit better the study of the factors relating to recruitment into OCGs.
190
T. Comunale et al.
Age
8
Gender
Ethnicity
Employment
6
2
3
5
5
Social ties
1
8
1
Economic conditions
3
4
4
4
Educational background
Group identity
6
4
14
4
7
27
1
17
Psychological factors
4
Criminal background and skills
4
2
6
25
Silence/Omertà
8
0%
10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Quantitative
Qualitative
Mixed-methods
Source: authors’ elaboration of the studies included in this systematic review.
Fig. 2 Factor categories per type of study (n = 184). (Source: authors’ elaboration of the studies
included in this systematic review.)
ocial, Psychological, and Economic Factors Relating
S
to Recruitment to Organized Crime Groups
Age
The reported age-groups of individuals involved into OC include both youngsters
(Arlacchi 1983; Arsovska 2015; Behan 1996; Blokland et al. 2017; Carvalho and
Soares 2016; Ciconte 1992; Cressey 1969; Hixon 2010; Kirby et al. 2016; Ostrosky
et al. 2012) and adults (Blokland et al. 2017; Kissner and Pyrooz 2009; Kleemans
and de Poot 2008; Salinas et al. 2011; Salinas and Regadera 2016; Unlu and Ekici
2012; Van Koppen et al. 2010b). Vulnerable young people join OCGs during adolescence and early adulthood. They are attracted by lifestyles dominated by the cult
of violence and by the show off of material goods (Behan 1996; Hixon 2010;
Ostrosky et al. 2012). The risk of recruitment at younger ages is also positively
associated with problems at school, drug use, low socio-economic conditions, and
the feeling of disillusion for the future (Arsovska 2015; Behan 1996; Carvalho and
Soares 2016). As a result, OC offenders are significantly younger than general
offenders in terms of age at first sanction (Kirby et al. 2016). The perspective of
profits, instead, primarily attracts adults. Social connections and skills acquired
with age facilitate their engagement in OC (Blokland et al. 2017; Kleemans and de
Poot 2008). These two age-groups emerge across the different types of OGCs, with
only few exceptions – outlaw motorcycle gangs specifically restrict membership to
adults (Blokland et al. 2017).
Systematic Review of the Social, Psychological and Economic Factors Relating…
191
Gender
The male population prevails across all the different types of OCGs (Brotherton and
Barrios 2004; Hixon 2010; Kirby et al. 2016; Salinas et al. 2011; Unlu and Ekici
2012; Varese 2011a, 2013). The rate of OC male offenders is even higher compared
with the population of general offenders (Salinas et al. 2011). The participation of
women occurs at lower rates mainly through family or emotional ties, especially
within DTOs (Brotherton and Barrios 2004; Requena et al. 2014; Salinas and
Regadera 2016; Van San and Sikkens 2017; Varese 2011a). Their role varies across
types of OCGs, and may change from home country to new territories.
Ethnicity
Criminological theories of strain, social disorganization, and subcultures developed
over the last century examined ethnic differences in offending and crime patterns.9
In line with these theories, ethnic marginality has been found to be associated with
involvement in OC. OCG membership help individuals to overcome the marginalization of their ethnic minority status (Arsovska 2015; Gordon 2000). Marginalization
can also help explaining the involvement in OC of illegal immigrants, whose irregular situation prevent them from finding a legitimate job (Salinas and Regadera
2016). Moreover, some ethnic groups are exposed to a higher risk of joining certain
types of OCGs, e.g. black individuals in drug-trafficking gangs (Carvalho and
Soares 2016).
Ethnic homogeneity proved to be relevant for most types of OCGs (Albini 1971;
Ciconte 1992; Cressey 1969; Decker and Chapman 2008; Hess 1993; Lupo 1993;
Varese 2011b, 2013; Wang 2013; Zhang and Chin 2002). Ethnic ties favor mutual
trust relations, which play a major role in OCGs due to the illicit nature of the activities and the risk of detection and arrest for their members (Decker and Chapman
2008; Zhang and Chin 2002). Ethnic homogeneity not only has favoured the establishment of OCGs within their territory of origin (Arlacchi 1983; Behan 1996;
Brancaccio 2017; Ciconte 1992; Gambetta 1993; Hess 1993), but has also contributed to their expansion across countries. With this regard, the literature reports on
Italian mafias in the U.S. and Germany, Russian mafias in Italy, and maras (i.e.
Central American street gangs) and Chinese triads in the U.S. (Albini 1971; Cressey
1969; Miguel Cruz 2010; Sciarrone and Storti 2014; Varese 2011a, 2013; Zhang and
Chin 2002).
Criminological theory states that ethnic heterogeneity coupled with low economic status, residential mobility, and family disruption lead to increases in crime rates among neighborhoods (Shaw
and McKay 1972). Subcultural theories explain crime involvement among lower-class individuals
belonging to ethnic minorities as a result of frustration and/or reaction to the dominant culture
(Cloward and Ohlin 1960; Cohen 1955; Miller 1958).
9
192
T. Comunale et al.
Educational Background
Low education levels are positively associated with individuals’ recruitment into
OCGs (Albini 1971; Carvalho and Soares 2016; Ciconte 1992; Hess 1993; Sales
2015). In the case of mafias, illiteracy does not prevent members from having a long
criminal career, becoming bosses and reaching wealth and status (Hess 1993).
However, mafia members’ educational levels may have increased over time
(Arlacchi 1983). While in the past mafia members were mainly illiterate and lower
class individuals, since the 1960s mafiosi include literate middle class individuals
(Arlacchi 1983). Nevertheless, the growth in mafia members’ education levels may
simply reflect the general increased education in the overall population during the
same period (Roser and Ortiz-Ospina 2017).
Employment
The lack of legitimate occupations and unemployment facilitates the involvement
into OCGs (Brancaccio 2017; Ciconte 1992; Gambetta 1993; Jhi and Gerber 2015;
Sales 2015; Wang 2013). A study has found the probability of becoming a gang
member almost three times lower for individuals with employment experience than
for those who have always been unemployed (Jhi and Gerber 2015). However, some
strategic occupations may also offer opportunities for criminal behaviour (Ciconte
1992; Kleemans and de Poot 2008; Kleemans and Van de Bunt 2008; Salinas et al.
2011; Van Koppen 2013). These occupations share some characteristics, as a certain
degree of independence to employees, the social exchange, and the relation with
mobility and logistics. Workers of the transport industry, especially truck drivers,
owners of car companies, and car dealers are attractive for OCGs (Kleemans and de
Poot 2008; Kleemans and Van de Bunt 2008; Van Koppen 2013). In these cases, licit
employments can conceal illicit activities, and illicit activities in turn can complement the legal salaries (Salinas et al. 2011).
Economic Conditions
Growing in poor city suburbs and socio-economically deprived environments facilitates recruitment into OCGs (Brotherton and Barrios 2004; Carvalho and Soares
2016; Ciconte 1992; Decker and Chapman 2008; Gordon 2000; Jhi and Gerber
2015; Sergi 2016; Van San and Sikkens 2017). Individuals from low socio-­economic
status may drop out of school to contribute to the family expenses and to improve
their living conditions. In such contexts, criminal activities may be attractive as a
way to pursue material and emotional reward (Gordon 2000; Ostrosky et al. 2012).
A career into Italian mafias, for example, fascinate many youngsters living in suburbs characterised by lack of money, goods, or means of support, contributing to
their continuous recruitment (Albini 1971; Arsovska 2015; Behan 1996; Brancaccio
2017; Hess 1993; Sales 2015). Nevertheless, individuals from higher socio-­
Systematic Review of the Social, Psychological and Economic Factors Relating…
193
economic conditions can also join OCGs, mainly attracted by high monetary gain
(Gordon 2000; Unlu and Ekici 2012; Van Koppen 2013; Wang 2013). This could be
a factor facilitating recruitment of individuals facing debts, specific needs (e.g. drug
addiction), or difficult life events (e.g. sudden illness) (Van Koppen 2013).
Social Ties
Social relations facilitate the involvement into criminal activities among the different types of OCGs (McIllwain 1999). Social relations may be developed in disparate spheres of everyday life (e.g. leisure and work ties), and later become a bridge
to OC (Decker and Chapman 2008; Gambetta 1993; Gordon 2000; Kleemans and
de Poot 2008; Kleemans and Van de Bunt 2008; Morselli 2003; Paoli 2003; Sales
2015; Van Koppen 2013; Van San and Sikkens 2017; Varese 2001, 2011b, 2013;
Wang 2013). This mechanism has been defined as “social opportunity structure”
(Kleemans and de Poot 2008). Individuals may join criminal settings because of
their experience and expert knowledge, and/or their ability to connect criminal
members with third parties, other criminal groups or representatives of the legal
businesses (Morselli 2003; Varese 2011a). Work, social, and criminal ties are therefore interwoven. Leisure activities and sidelines often act as a bridge between representatives of the legal and criminal settings (Kleemans and de Poot 2008;
Kleemans and Van de Bunt 2008; Van Koppen 2013). The social opportunity structure may explain why some individuals get involved into OCGs during adulthood:
their relational network enables their expertise to match a specific criminal opportunity. Moreover, individuals’ attitude towards criminal opportunities may loosen due
to improvise life events (e.g., bankruptcy) (Kleemans and de Poot 2008; Van
Koppen 2013).
Kinship and blood ties are facilitators to access OCGs (Arlacchi 1983; Arsovska
2015; Behan 1996; Brancaccio 2017; Ciconte 1992; Cressey 1969; Decker and
Chapman 2008; Gambetta 1993; Kissner and Pyrooz 2009; Lo 2010; Requena et al.
2014; Salinas et al. 2011; Salinas and Regadera 2016; Sciarrone 2014; Sergi 2016;
Van Koppen 2013; Varese 2001, 2006, 2011a). The key role of kinship and blood
ties in the recruitment into OCGs results from the need of cultural transmission and
gradual learning of criminal culture and knowledge to new members (Gordon 2000;
Sergi 2016). Family members and close friends are easier to trust in the criminal
environment lacking third parties to regulate businesses and disputes (Van Koppen
2013). Being part of a criminal family may per se facilitate one’s recruitment into
OCGs, especially in the mafias (Albini 1971; Arlacchi 1983; Ciconte 1992;
Gambetta 1993; Hess 1993; Lupo 1993; Paoli 2003; Sciarrone 2014; Sergi 2016).
In the mafias, the concept of family goes beyond proper kinship and blood ties.
Individuals can assume a permanent new identity as “men of honor” in OCGs
through the acceptance of a “status contract”, through their affiliation ceremony
(which in some Italian mafias is called “baptism”). During this ceremony, several
rituals help creating brotherhood ties among members, thus also enabling “contract
of fraternization”. This contract extends the family bonds of loyalty and obligation,
194
T. Comunale et al.
providing members with the mutual support and trust necessary to pursue groups’
aims (Paoli 2003). Family ties play a significant role also for gangs, in which parental gang membership has been proven to be a significant predictor of individuals’
affiliation (Kissner and Pyrooz 2009).
Group Identity
The sense of belonging and social identity associated with the membership into a
criminal group facilitates the recruitment into OCGs (Arsovska 2015; Brotherton
and Barrios 2004; Densley 2012; García 2006; Hixon 2010; May 2009). Co-offending
and social interactions within the OCGs reinforce the sense of loyalty and social
cohesion among members (Densley 2012). These processes enhance members’
sense of group identity (García 2006). Moreover, individuals joining OCGs in many
cases belong to subcultures characterized by values of honor and loyalty, and are
often recruited through ritualized procedures of affiliation (Albini 1971; Arlacchi
1983; Brotherton and Barrios 2004; Cressey 1969; Gambetta 1993; Hess 1993;
Hixon 2010; Lo 2010; Paoli 2003; Sciarrone 2014; Sergi 2016; Zhang and Chin
2002). This is particularly relevant in the case of mafias, in which the initiation ceremonies play a key role (Albini 1971; Gambetta 1993; Hess 1993; Hixon 2010;
Paoli 2003; Sergi 2016). To be recruited into the mafias, individuals must have psychic attitudes and a moral code typical of “men of honor”, who are respected and
capable of violent revenge if necessary (Gambetta 1993; Hess 1993). During the
initiation ceremonies, the new members have to take an oath of “providing for the
family of a member in the event of his death or incarceration” (Albini 1971,
pp. 113–114; see also Paoli 2003; Sergi 2016).
Psychological Factors
Some studies associate substance use disorder, childhood conduct disorder, and/or
abnormal psychopathological traits to an increase of individuals’ risk of involvement in OC activities (Hixon 2010; Kissner and Pyrooz 2009; May 2009; Ostrosky
et al. 2012; Schimmenti et al. 2014; Sergi 2016). Individuals recruited into OCGs
usually have an extensive history of negative and arrested development during adolescence, leading to antisocial personality disorders during their adulthood
(Hixon 2010).
A medium–high psychopathy level is common among most OC members, in line
with their criminal skills. OC offenders lack empathy and inhibition, they have
marked cruelty and callousness, affective impairments, and an average antisocial
lifestyle. Such profiles are compatible with violent and risky activities, and committing physical assaults for immediate gratification, regardless of potential prosecution (Ostrosky et al. 2012). Another study confirmed the psychopathy and antisocial
traits characterizing mafias’ members, also underlying their infantile and dependent
personality traits, demonstrated by their obedience to the willingness of their orga-
Systematic Review of the Social, Psychological and Economic Factors Relating…
195
nizations. Mafia members differ from other offenders in that they do not suffer from
substance use disorder. Substance use may indeed impair the reliability of mafia
members, who always have to display the honorable principles of their organizations (Schimmenti et al. 2014).
Nevertheless, criminals involved in “high-profile” activities, like money-­
laundering, present different characteristics. These individuals maintain quite a normal lifestyle, but they lack affective bonds, they have increased arrogance,
callousness, and a strong desire for recognition and for a certain economic status.
These features allow them to achieve their goals through unethical and illicit nonviolent activities (Ostrosky et al. 2012).
Criminal Background and Skills
Attitudes towards violent and risk-taking behaviors are essential to be part of
OCGs. Violence is key to reinforce the criminal status of members, obtaining the
respect and social acceptance of their group (Albini 1971; Behan 1996).
Consequently, people with a criminal background are more likely to become
involved in OCGs (Albini 1971; Arlacchi 1983; Behan 1996; Blokland et al. 2017;
Brancaccio 2017; Gambetta 1993; Hess 1993; Kirby et al. 2016; Requena et al.
2014). In many cases OC members have a long criminal history (Albini 1971;
Arsovska 2015; Blokland et al. 2017; Brancaccio 2017; Cressey 1969; Decker
and Chapman 2008; Densley 2012; Gambetta 1993; Kirby et al. 2016; Kleemans
and de Poot 2008; Lupo 1993; Morselli 2003; Sales 2015; Salinas et al. 2011; Van
Koppen et al. 2010a, b; Varese 2011b). Compared to the population of general
offenders, OC offenders show higher rates of offending seriousness, and serve
longer time in prison (Van Koppen et al. 2010a). The key role of the cult of violence also explains the positive association between prison background and
recruitment into OCGs, as individuals seek for protection and personal safety
within the penitentiary (Behan 1996; Ciconte 1992).
The possession of criminal skills or special expertise is another relevant feature for recruitment into OCGs. The former refers to the ability of avoiding
police detection and allows OCGs to test the loyalty and criminal potential of
individuals (Densley 2012). The latter refers to high levels of knowledge, or the
right set of skills, making individuals suited for certain types of illegal businesses. In some cases, the special expertise required can be very specific, as for
translators, people capable of handling explosives, or people with a chemistry
background supporting drug-related operations (Gambetta 1993). Such special
expertise can derive from criminal or legitimate experience (Van Koppen 2013;
Salinas and Regadera 2016). In both cases, having a special expertise explains
the involvement in OCGs during adulthood (Kleemans and de Poot 2008).
Individuals with specific criminal expertise acquired through their criminal
career may be recruited in prison (Behan 1996; Ciconte 1992) or outside
(Arlacchi 1983; Cressey 1969; Gambetta 1993; Lo 2010; Salinas et al. 2011;
Salinas and Regadera 2016; Sergi 2016; Van Koppen 2013).
196
T. Comunale et al.
Silence/Omertà
Individuals joining OCGs cannot break the code of silence, also known as omertà,
required to commit OC activities (Albini 1971; Ciconte 1992; Cressey 1969;
Gambetta 1993; Hess 1993; Lo 2010; Lupo 1993; Paoli 2003; Requena et al. 2014).
For mafia members, handling secrecy also implies the impossibility to have contacts
or any relationships with legal authorities. Albini (1971, pp. 107–108) underlines
that “omertà is not unique to any one organization or society. Omertà is a behavior
and attitude [and] […] it represents suspicion and resentment of government
and law”.
Discussion
This systematic review includes 47 studies published over a span of nearly fifty
years (1969–2017). Results show that the majority of the studies employs qualitative methods and cover the entire period, though quantitative research has recently
increased. Compared to the previous decade (i.e. 2000–2009), literature on recruitment into OCGs has risen by almost 108%, with 27 included studies published since
2010. The growth has encompassed all type of studies, but the most notable increase
has involved the use of quantitative methods (from 2 studies during 2000–2009 to
10 since 2010).
As for the geographic scope, the recent rising trend has mostly occurred in
Europe. Of the total number of included studies, almost 40% (n = 18) investigate
recruitment into OCGs in the European context and were published on or after
2010. Italy is the most represented country, though academic studies on the
Netherlands and Spain have also emerged over the last two decades. After Europe,
the most represented continent is North America, followed by a small number of
comparative research studies (e.g. OC in the U.S. and China).
Results therefore show that literature focuses on Europe rather than the Americas,
Russia, and Far East, other geographic areas also affected by the presence of OCGs.
This finding suggests that the interest in OC-related activities is prevalent in some
countries, while in others there seems to be less attention despite a long tradition in
criminological research (e.g. the U.S.). Nonetheless, the advancement of quantitative research in the study of OC may change this scenario, fostering the interest of
more scholars. Further quantitative studies in under-represented countries would
contribute to enhance the knowledge on risk factors for involvement into OCGs.
Research findings show that factors associated with recruitment into OCGs are
highly interrelated. Social and economic factors play a major role while psychological
factors are found to be less important. This finding is in line with the literature on the
social opportunity structure (Kleemans and De Poot 2008) as the underlying mechanism
through which recruitment into OCGs occurs. Opportunities to enter OCGs depend on
individuals’ personal background and skills, as well as on the social relations they
develop in different settings of their lives. Previous studies indeed consistently highlight
Systematic Review of the Social, Psychological and Economic Factors Relating…
197
the importance of factors as violent attitudes and behavior, criminal background, low
economic status, and social relations in facilitating individuals’ involvement into
OC. Members of OCGs usually have a significant criminal history and/or a prison background resulting from their proneness to violent and risk-taking behavior. This is particularly true for people living in poor and socially disorganized areas. The lack of access
to legitimate means to attain commonly accepted goals may make them more prone to
seek illegal alternatives to overcome difficulties. In this perspective emerges the social
embeddedness of OC. Recruitment in OCGs mainly revolves on social proximity and
interaction with members of criminal organizations. Although relations developed in
work settings and everyday activities are important in recruiting new members, kinship
and blood ties with OC offenders allow for increased mutual trust and loyalty.
Though social, psychological, and economic factors are highly interrelated concerning criminal recruitment, some are equally present across various types of
OCGs, while others are particularly relevant for some of them. Kinship and blood
ties are particularly important in recruiting mafia members. Furthermore, these
groups succeed in extending the family concept beyond proper kinship and blood
ties. Through the initiation ritual, individuals accept the “contract of fraternization”
enhancing brotherhood ties among members, and increasing the cohesion of the
criminal organizations and reinforces group identity. This latter aspect is a key factor also for gang membership. Rituals and a common culture provide individuals
with a sense of belonging that favors their involvement in gangs. Conversely, individuals usually join DTOs mainly driven by monetary returns. Finally, individuals
get involved in other criminal organizations during adulthood as they develop specialized expertise and social ties with criminals only later in life.
The identification of a broad variety of factors on recruitment into OCGs benefits
from studies adopting different methodological approaches. Included studies are consistent in pointing to the same factors, regardless of their quantitative, qualitative, or
mixed-methods approach. Nonetheless, each approach better suits the analysis of specific factors. For instance, while quantitative studies usually report on age, gender, and
economic conditions, qualitative research often deals with factors that can hardly be
quantified, as group identity and silence/omertà (Fig. 2). Relevant grey literature further corroborates the consistency of the results. Selected reports identify the same
factors emerging from the review on empirical studies on the Italian mafias operating
in their territory of origin, as well as on foreign mafias operating in other geographic
areas. However, the inclusion of studies adopting d­ ifferent methodological approaches
(Fig. 1), as well as the heterogeneity of the concept of OC, may limit the generalizability of research results beyond the OCG and the geographic area under analysis in
each study. On the one hand, the main limitations of included qualitative studies relate
to their research designs. Many of them rely on case studies, thus lacking external
validity. Alternatively, they are monographies focusing on the mafias in their territories of origin. On the other hand, the main limitation of quantitative and mixed-methods studies lies in their data sources. They mostly rely on investigative and judicial
data, thus referring to non-random samples that suffer from the “dark number” issue
(i.e. the discrepancy between the number of real events and the reported offences and
offenders). Notwithstanding the methodological approach, the limited generalizabil-
198
T. Comunale et al.
ity of the results may be due to the heterogeneity of the concept of OC. Given the lack
of a unique definition of OC, the literature in this field of study encompasses varying
regions and criminal organizations. The adoption of a broad definition of OC (i.e. by
identifying four types of OCGs) allows providing a general overview of the factors
relating to recruitment into OC, although results should be interpreted with caution
when considering other types of OCGs. Nonetheless, the results of this systematic
review provide a systematized and comprehensive understanding on the factors facilitating pathways into OC, while also pointing out relatively under-examined factors for
future studies. Finally, research findings contribute to consolidate the knowledge in
this field and may serve to inform the policy-making process aimed at preventing
involvement into OC.
Conclusions
This study is the first systematic review shedding light on the most commonly
reported social, psychological, and economic factors associated with involvement
into OC. The extensive and rigorous search – performed in multiple languages and
several databases – has allowed identifying and assessing potentially eligible
records while also reducing the risk of bias in the inclusion of relevant literature.
The systematic review includes 47 studies, of which the majority are qualitative
(57%), followed by quantitative studies (26%) and mixed-methods ones (17%). The
literature includes different types of OCGs – among all Italian mafias (40% of the
studies) and other criminal organizations (30%) – with official statistics, investigative and judicial files as the main source of data.
Research findings highlight that social and economic factors play a major role for
involvement into OC. Individuals’ recruitment into OCGs is influenced, among all, by
low socio-economic status, attraction for a strong group identity, and social ties. Social
connections and work-related ties drive the recruitment of late-onset offenders, who
get involved into OC because of criminal opportunities, which may arise at later stages
in life. Financial gains are also found to be important, especially for individuals who
join DTOs. An existing criminal background or possession special expertise constitute
a positive driver for involvement into OC, although some OCGs as the mafias require
peculiar skills as the ability of handling secrecy and enforce the code of silence.
As for limitations, the research design adopted by qualitative and mixed-­methods
approaches might mitigate generalizability of findings to other OCGs or other geographic areas. Nonetheless, this study provides a more comprehensive understanding on factors associated with involvement in OC and contributes to the consolidation
of such knowledge also providing insights on the empirical evidence by geographic
areas and types of literature included (i.e. types of methods and data sources). In
addition, the heterogeneity of the concept of OC has made it difficult to perform a
meta-analysis to properly combine and analyze the results from multiple quantitative studies. Despite these limitations, this systematic review is the first study to
synthesize the empirical evidence on factors related to recruitment into OCGs.
Systematic Review of the Social, Psychological and Economic Factors Relating…
199
Future studies may expand and improve this review under some aspects. One of
the main implications for future research regards the methodological approach in
the study of recruitment into OC. If the recently emerged quantitative approach proliferates, results of future studies will have an increased external validity. Quantitative
research will also allow the development and testing of hypotheses and therefore the
empirical test of theoretical contributions in the field of OC. In addition, further
quantitative research will allow further enhancing the empirical evidence on risk
factors for involvement into OC and comparing such dynamics with general or less
serious crimes. These improvements could be feasible provided the exchange of
data and information among law enforcement agencies, scholars, and policy-­
makers. Increasing publication of quantitative studies would also facilitate the comparison of results across different types of OCGs and regions. Future studies may
adopt different scopes, focusing on different types of OCGs or on OCGs active in
specific geographic areas that are currently underrepresented (e.g. Asian or African
countries). Overall, scholars should focus on factors relating to recruitment into
OCGs with more detail and more comprehensively. They may therefore focus on
both those factors emerging as crucial (e.g. individuals’ criminal background and
social relations) and others that are currently underreported (e.g. psychological factors as personality traits).
These improvements would positively influence future OC research as well as
practitioners and policy-makers’ work. Extensive and systematic knowledge of the
factors relating to recruitment into OCGs would allow institutions to tackle this
continuous process. The analysis identification of individual red flags combined
with the assessment of risky social environments would facilitate the design of tailored policies aimed at preventing recruitment into OCGs. The limited generalizability of the research findings prevents the identification of specific recommendations
for crime prevention policies. Nonetheless, the results of this systematic review can
inform future policy-making processes enhanced by further empirical research on
OC and involvement into OCGs.
References
Studies Included in the Systematic Review
Albini, J. L. (1971). The American mafia: Genesis of a legend. New York: Appleton-Century-Crofts.
Arlacchi, P. (1983). La mafia imprenditrice. Bologna: Il Mulino.
Arsovska, J. (2015). Decoding Albanian organized crime: Culture, politics, and globalization.
Oakland: University of California Press.
Behan, T. (1996). The camorra. London: Routledge.
Blokland, A., van Hout, L., van der Leest, W., & Soudijn, M. (2017). Not your average biker;
criminal careers of members of Dutch outlaw motorcycle gangs. Trends in Organized Crime.,
22, 10–33. https://doi.org/10.1007/s12117-017-9303-x.
Brancaccio, L. (2017). I clan di camorra: genesi e storia. Roma: Donzelli Editore.
200
T. Comunale et al.
Brotherton, D., & Barrios, L. (2004). The almighty Latin King and Queen Nation: Street politics
and the transformation of a New York City gang. New York/Chichester: Columbia University
Press.
Carvalho, L. S., & Soares, R. R. (2016). Living on the edge: Youth entry, career and exit in
drug-selling gangs. Journal of Economic Behavior & Organization, 121, 77–98. https://doi.
org/10.1016/j.jebo.2015.10.018.
Ciconte, E. (1992). “Ndrangheta dall”unità a oggi. Bari: Editori Laterza.
Cloward, R., & Ohlin, L. (1960). Delinquency and Opportunity. New York: The Free Press.
Cressey, D. R. (1969). Theft of the nation: The structure and operations of organized crime in
America. New Brunswick/London: Transaction Publishers.
Cohen, A. K. (1955). Delinquent Boys: The Culture of the Gang. Glencoe, IL: Free Press.
Decker, S. H., & Chapman, M. T. (2008). Roles, recruitment into, and remaining involved in
the drug smuggling trade. In Drug smugglers on drug smuggling: Lessons from the inside
(pp. 88–113). Philadelphia: Temple University Press.
Densley, J. A. (2012). Street gang recruitment: Signaling, screening, and selection. Social
Problems, 59(3), 301–321. https://doi.org/10.1525/sp.2012.59.3.301.
Gambetta, D. (1993). The Sicilian mafia: The business of private protection. Cambridge, MA:
Harvard University Press.
García, M. M. (2006). “Narcoballads”: The psychology and recruitment process of the “Narco.”.
Global Crime, 7(2), 200–213. https://doi.org/10.1080/17440570601014461.
Gordon, R. M. (2000). Criminal business organizations, street gangs and wanna-be groups: A
Vancouver perspective. Canadian Journal of Criminology, 42, 39–60.
Hess, H. (1993). Mafia: Le origini e la struttura. Roma-Bari: Editori Laterza.
Hixon, S. J. (2010). Archetypal perspectives on Nordic and Germanic initiation symbols, mythology, and rites of passage in a European American self-referenced White supremacist gang (PhD
dissertation, Saybrook University). Ann Arbor: ProQuest/UMI (Publication No. 3418929).
Jhi, K. Y., & Gerber, J. (2015). Texan gangs in “Da Hood:” The impact of actual and perceptual
neighborhood qualities on gang membership. Justice Policy Journal, 12, 1–25.
Kirby, S., Francis, B., Humphreys, L., & Soothill, K. (2016). Using the UK general offender database as a means to measure and analyse organized crime. Policing: An International Journal,
39(1), 78–94. https://doi.org/10.1108/PIJPSM-03-2015-0024.
Kissner, J., & Pyrooz, D. C. (2009). Self-control, differential association, and gang membership:
A theoretical and empirical extension of the literature. Journal of Criminal Justice, 37(5), 478–
487. https://doi.org/10.1016/j.jcrimjus.2009.07.008.
Kleemans, E. R., & de Poot, C. J. (2008). Criminal careers in organized crime and social
opportunity structure. European Journal of Criminology, 5(1), 69–98. https://doi.
org/10.1177/1477370807084225.
Kleemans, E. R., & Van de Bunt, H. G. (2008). Organised crime, occupations and opportunity.
Global Crime, 9(3), 185–197. https://doi.org/10.1080/17440570802254254.
Lo, W. T. (2010). Beyond social capital: Triad organized crime in Hong Kong and China. British
Journal of Criminology, 50(5), 851–872. https://doi.org/10.1093/bjc/azq022.
Lupo, S. (1993). Storia della mafia. Roma: Donzelli Editore.
May, C. (2009). The relationship between risk factors, social support and gangs (PhD dissertation,
Alliant International University). Ann Arbor: ProQuest/UMI (Publication No. 3421105).
McIllwain, J. S. (1999). Organized crime: A social network approach. Crime, Law and Social
Change, 32(4), 301–323. https://doi.org/10.1023/A:1008354713842.
Miguel Cruz, J. (2010). Central American maras: From youth street gangs to transnational protection rackets. Global Crime, 11(4), 379–398. https://doi.org/10.1080/17440572.2010.519518.
Miller, W. B. (1958). Lower Class Culture as a Generating Milieu of Gang Delinquency. Journal of
Social Issues, 14(3), 5–19. https://doi.org/10.1111/j.1540-4560.1958.tb01413.x.
Morselli, C. (2003). Career opportunities and network-based privileges in the Cosa Nostra. Crime,
Law and Social Change, 39(4), 383–418. https://doi.org/10.1023/A:1024020609694.
Systematic Review of the Social, Psychological and Economic Factors Relating…
201
Ostrosky, F., Borja, K. C., Rebollar, C. R., & Díaz, K. X. (2012). Neuropsychological profiles of
members of organized crime and drug-traffic organizations. Research and Reports in Forensic
Medical Science, 19–30. https://doi.org/10.2147/RRFMS.S32352.
Paoli, L. (2003). Mafia brotherhoods: Organized crime, Italian style. Oxford: Oxford University
Press.
Requena, L., de Juan, M., Salinas, A. G., & de la Corte, L. (2014). A psychosocial study on crime
and gender: Position, role and status of women in a sample of Spanish criminal organizations /
Un estudio psicosocial sobre la delincuencia y género. Posición, rol y estatus de la mujer en una
muestra española de organizaciones criminales. Revista de Psicología Social, 29(1), 121–149.
https://doi.org/10.1080/02134748.2013.878572.
Roser, M., & Ortiz-Ospina, E. (2017). Global Education. Published online at OurWorldInData.
org. Retrieved from: https://ourworldindata.org/global-rise-of-education [Online Resource]
Sales, I. (2015). Storia dell’Italia mafiosa: perché le mafie hanno avuto successo. Soveria
Mannelli: Rubbettino.
Salinas, A. G., & Regadera, S. F. (2016). Multiple affiliations in criminal organizations: Analysis
of a Spanish sample. Crime, Law and Social Change, 65(1–2), 47–65. https://doi.org/10.1007/
s10611-015-9597-z.
Salinas, A. G., Requena, L., & De la Corte, L. (2011). ¿Existe un perfil de delincuente organizado? Exploración a partir de una muestra española. Revista Electrónica de Ciencia Penal Y
Criminología, 13(3), 03:1–03:32.
Schimmenti, A., Caprì, C., La Barbera, D., & Caretti, V. (2014). Mafia and psychopathy. Criminal
Behaviour and Mental Health, 24(5), 321–331. https://doi.org/10.1002/cbm.1902.
Sciarrone, R. (2014). `Ndrangheta: A reticular organization. In N. Serenata (Ed.), The
`Ndrangheta and Sacra Corona Unita (Vol. 12, pp. 81–99). Cham: Springer. https://doi.
org/10.1007/978-3-319-04930-4_6.
Sciarrone, R., & Storti, L. (2014). The territorial expansion of mafia-type organized crime. The
case of the Italian mafia in Germany. Crime, Law and Social Change, 61(1), 37–60. https://doi.
org/10.1007/s10611-013-9473-7.
Sergi, A. (2016). A qualitative reading of the ecological (dis)organisation of criminal associations.
The case of the “Famiglia Basilischi” in Italy. Trends in Organized Crime, 19(2), 149–174.
https://doi.org/10.1007/s12117-015-9254-z.
Shaw, C. R., & McKay, H. D. (1972). Juvenile delinquency and urban areas. Chicago: University
of Chicago Press.
Unlu, A., & Ekici, B. (2012). The extent to which demographic characteristics determine international drug couriers’ profiles: A cross-sectional study in Istanbul. Trends in Organized Crime,
15(4), 296–312. https://doi.org/10.1007/s12117-012-9152-6.
Van Koppen, M. V. (2013). Involvement mechanisms for organized crime. Crime, Law and Social
Change, 59(1), 1–20. https://doi.org/10.1007/s10611-012-9396-8.
Van Koppen, M. V., de Poot, C. J., & Blokland, A. A. J. (2010a). Comparing criminal careers
of organized crime offenders and general offenders. European Journal of Criminology, 7(5),
356–374. https://doi.org/10.1177/1477370810373730.
Van Koppen, M. V., de Poot, C. J., Kleemans, E. R., & Nieuwbeerta, P. (2010b). Criminal trajectories in organized crime. British Journal of Criminology, 50(1), 102–123. https://doi.
org/10.1093/bjc/azp067.
Van San, M., & Sikkens, E. (2017). Families, lovers, and friends: Women, social networks, and
transnational cocaine smuggling from Curaçao and Peru. The Howard Journal of Crime and
Justice, 56(3), 343–357. https://doi.org/10.1111/hojo.12218.
Varese, F. (2001). The Russian mafia: Private protection in a new market economy. Oxford/New
York: OUP Oxford.
Varese, F. (2006). How mafias migrate: The case of the `Ndrangheta in northern Italy. Law &
Society Review, 40(2), 411–444. https://doi.org/10.1111/j.1540-5893.2006.00260.x.
Varese, F. (2011a). How mafias take advantage of globalization: The Russian mafia in Italy. British
Journal of Criminology, 52(2), 235–253. https://doi.org/10.1093/bjc/azr077.
202
T. Comunale et al.
Varese, F. (2011b). Mafias on the move: How organized crime conquers new territories. Princeton:
Princeton University Press.
Varese, F. (2013). The structure and the content of criminal connections: The Russian mafia in
Italy. European Sociological Review, 29(5), 899–909. https://doi.org/10.1093/esr/jcs067.
Wang, W. (2013). Conventional capital, criminal capital, and criminal careers in drug trafficking
(PhD dissertation, Arts and Social Sciences: School of Criminology). Burnaby, B.C.: Simon
Fraser University Library.
Zhang, S., & Chin, K.-L. (2002). Enter the dragon: Inside Chinese human smuggling organizations. Criminology, 40(4), 737–768. https://doi.org/10.1111/j.1745-9125.2002.tb00972.x.
Other References
Albini, J. L. (1971). The American mafia: Genesis of a legend. New York: Appleton-Century-Crofts.
Albrecht, H.-J., & Fijnaut, C. (Eds.). (2002). The containment of transnational organized crime:
comments on the UN convention of December 2000. Freiburg im Breisgau: Ed. iuscrim Max-­
Planck-­Inst. für Ausländisches und Internat. Strafrecht.
Bandiera, O. (2003). Land reform, the market for protection, and the origins of the Sicilian mafia:
Theory and evidence. Journal of Law, Economics, and Organization, 19(1), 218–244. https://
doi.org/10.1093/jleo/19.1.218.
Benzies, K. M., Premji, S., Hayden, K. A., & Serrett, K. (2006). State-of-the-evidence reviews:
Advantages and challenges of including grey literature. Worldviews on Evidence-Based
Nursing, 3(2), 55–61. https://doi.org/10.1111/j.1741-6787.2006.00051.x.
Blevins, K. R., Johnson Listwan, S., Cullen, F. T., & Lero Jonson, C. (2010). A general strain
theory of prison violence and misconduct: An integrated model of inmate behavior. Journal of
Contemporary Criminal Justice, 26(2), 148–166. https://doi.org/10.1177/1043986209359369.
Carvalho, L. S., & Soares, R. R. (2016). Living on the edge: Youth entry, career and exit in
drug-selling gangs. Journal of Economic Behavior & Organization, 121, 77–98. https://doi.
org/10.1016/j.jebo.2015.10.018.
Clapper, J. R. (2015). Statement for the record worldwide threat assessment of the U.S. IC
before the Senate Armed Services Committee. Washington: Office of the Director of National
Intelligence. Retrieved from https://www.dni.gov/files/documents/Unclassified_2015_ATA_
SFR_-_SASC_FINAL.pdf.
Cornish, D. B., & Clarke, R. V. (2002). Analyzing organized crime. In A. R. Piquero & S. G.
Tibbetts (Eds.), Rational choice and criminal behavior: Recent research and future challenges.
New York: Garland.
Cressey, D. R. (1969). Theft of the nation: The structure and operations of organized crime in
America. Transaction Publishers.
Critical Appraisal Skills Programme. (2017). CASP qualitative checklist. [online] Available at:
http://www.casp-uk.net/checklists Accessed: 11/10/2017.
Daniele, V. (2009). Organized crime and regional development. A review of the Italian case. Trends
in Organized Crime, 12(3–4), 211–234. https://doi.org/10.1007/s12117-009-9079-8.
Decker, S. H., & Pyrooz, D. C. (2014). Gangs another form of organized crime? In L. Paoli (Ed.),
Oxford handbook of organized crime. Oxford: Oxford University Press.
Dixon-Woods, M., Bonas, S., Booth, A., Jones, D. R., Miller, T., Sutton, A. J., et al. (2006). How
can systematic reviews incorporate qualitative research? A critical perspective. Qualitative
Research, 6(1), 27–44. https://doi.org/10.1177/1468794106058867.
Finckenauer, J. O. (2007). Mafia and organized crime: A beginner’s guide. Oxford: Oneworld
Publications.
Systematic Review of the Social, Psychological and Economic Factors Relating…
203
Fisher, H., Montgomery, P., & Gardner, F. (2008a). Cognitive-behavioural interventions for preventing youth gang involvement for children and young people. Campbell Systematic Reviews,
7, 1–35. https://doi.org/10.4073/csr.2008.7.
Fisher, H., Montgomery, P., & Gardner, F. (2008b). Opportunities provision for preventing youth
gang involvement for Children and Young People (7-16). Campbell Systematic Reviews, 8,
1–32. https://doi.org/10.4073/csr.2008.8.
Gambetta, D. (1993). The Sicilian mafia: The business of private protection. Harvard University
Press.
Hagan, F. E. (2015). Introduction to criminology: Theories, methods, and criminal behavior. Los
Angeles: SAGE Publications.
Hess, H. (1993). Mafia: Le origini e la struttura. Editori Laterza: Roma-Bari.
Higginson, A., Mazerolle, L., Benier, K. H., & Bedford, L. (2014). Predictors of youth gang
membership in low-and middle-income countries: A systematic review. Campbell Systematic
Reviews. Retrieved from https://campbellcollaboration.org/library/predictors-of-youth-gangmembership-low-and-middle-income-countries.html
Hodgkinson, J., Marshall, S., Berry, G., Newman, M., Reynolds, P., Burton, E., et al. (2009).
Reducing gang related crime: A systematic review of ‘comprehensive’ interventions. Technical
report. In Research evidence in education library. London: EPPI-Centre, Social Science
Research Unit, Institute of Education, University of London.
Ianni, F. A. J. (1974). Black mafia: Ethnic succession in organized crime (1st ed.). New York:
Simon & Schuster.
Jones, O. D., Wagner, A. D., Faigman, D. L., & Raichle, M. E. (2013). Neuroscientists in court.
Nature Reviews Neuroscience, 14(10), 730–736. https://doi.org/10.1038/nrn3585.
Kleemans, E. R., & De Poot, C. J. (2008b). Criminal careers in organized crime and social
opportunity structure. European Journal of Criminology, 5(1), 69–98. https://doi.
org/10.1177/1477370807084225.
Kleemans, E. R., & Van de Bunt, H. G. (1999). The social embeddedness of organized crime.
Transnational Organized Crime, 5(1), 19–36.
Kleemans, E. R., & Van de Bunt, H. G. (2008). Organised crime, occupations and opportunity.
Global Crime, 9(3), 185–197. https://doi.org/10.1080/17440570802254254.
Kleemans, E. R., & Van Koppen, M. V. (2014). Careers in organized crime. In G. Bruinsma
& D. Weisburd (Eds.), Encyclopedia of criminology and criminal justice (pp. 285–295).
New York: Springer.
Klein, M. W., & Maxson, C. L. (2006). Street gang patterns and policies. New York: Oxford
University Press.
Knox, J., & Gray, D. H. (2014). The national and international threat of drug trafficking organizations. Global Security Studies, 5(3), 27–34.
Langston, M. (2003). Addressing the need for a uniform definition of gang-involved crime. FBI
Law Enforcement Bulletin, 72, 7.
Lavezzi, A. M. (2008). Economic structure and vulnerability to organised crime: Evidence from
Sicily. Global Crime, 9(3), 198–220. https://doi.org/10.1080/17440570802254312.
Lavezzi, A. M. (2014). Organised crime and the economy: A framework for policy prescriptions.
Global Crime, 15(1–2), 164–190. https://doi.org/10.1080/17440572.2013.868626.
Murray, J., Farrington, D. P., & Eisner, M. P. (2009). Drawing conclusions about causes from
systematic reviews of risk factors: The Cambridge Quality Checklists. Journal of Experimental
Criminology, 5(1), 1–23. https://doi.org/10.1007/s11292-008-9066-0.
Ostrosky, F., Borja, K. C., Rebollar, C. R., & Díaz, K. X. (2012). Neuropsychological profiles of
members of organized crime and drug-traffic organizations. Research and Reports in Forensic
Medical Science, 19–30. https://doi.org/10.2147/RRFMS.S32352.
Paoli, L. (2014a). Organized crime, types of. In G. Bruinsma & D. Weisburd (Eds.), Encyclopedia
of criminology and criminal justice (pp. 3376–3387). New York: Springer.
Paoli, L. (2014b). The Italian mafia. In L. Paoli (Ed.), The Oxford handbook of organized crime
(pp. 121–141). New York: Oxford University Press.
204
T. Comunale et al.
U.S. Department of Justice. (2010). National drug threat assessment 2010 (National Drug
Intelligence Center). Retrieved from https://www.justice.gov/archive/ndic/pubs38/38661/
United Nations. (2000). Convention on transnational organized crime (No. Assembly Resolution
55/95 of 15 Nov 2000). Retrieved from http://www.unodc.org/unodc/treaties/CTOC/
Varese, F. (2010). Organized crime: Critical concepts in criminology. London/New York:
Routledge.
Walsh, A., & Beaver, K. M. (2009). Biosocial Criminology. In Handbook on crime and deviance.
New York: Springer.
Wood, J. L., Alleyne, E., Mozova, K., & James, M. (2014). Predicting involvement in prison gang
activity: Street gang membership, social and psychological factors. Law and Human Behavior,
38(3), 203–211. https://doi.org/10.1037/lhb0000053.
Socio-Economic Inequalities
and Organized Crime: An Empirical
Analysis
Michele Battisti, Giovanni Bernardo, Antri Konstantinidi,
Andros Kourtellos, and Andrea Mario Lavezzi
Introduction
In this chapter we contribute to the recent literature (e.g., Istat 2010; Acciari et al.
2017; Güell et al. 2017) that provides evidence that inequality is high and social
mobility is low in the Italian regions and provinces where organized crime is widespread such as those of Southern Italy. We complement this line of work in two
respects. First, using a novel panel dataset at the regional level for the period
1985–2014 we investigate the relationship between inequality and organized crime
at the regional level, exploiting both time and cross-sectional variation.1 Second, we
assess the role of social mobility in organized crime.
Our main hypothesis is that both inequality and intergenerational persistence
lead to organized crime. There exist various mechanisms for why income inequality
matters. In particular, direct channels may be at work, for example if poor individuals find organized crime attractive for lack of remunerative alternatives in the labor
market. Alternatively, rich individuals, for example members of the economic and
political élite, may demand organized crime “services” to gain or preserve their
Some previous work has studied the relationship between inequality and corruption (e.g., JongSung and Khagram 2005), violent crime (e.g. Fajnzylber et al. 2002) and civil conflict (e.g.,
Macours 2010).
1
M. Battisti · G. Bernardo · A. M. Lavezzi (*)
Department of Law, University of Palermo, Palermo, Italy
e-mail: mario.lavezzi@unipa.it
A. Konstantinidi
Department of Economics, University of Cyprus, Nicosia, Cyprus
Cyprus and Department of Law, University of Palermo, Palermo, Italy
A. Kourtellos
Department of Economics, University of Cyprus, Nicosia, Cyprus
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_9
205
206
M. Battisti et al.
socio-economic standing. Franchetti (1877) initially pointed out this behavior in his
thorough enquiry on Sicily at the time of reunification of Italy, as he noted that the
Sicilian upper class benefited from the protection services by the early mafiosi.
Recent evidence supporting this claim is found for example in Gambetta and Reuter
(1995), describing how entrepreneurs can gain monopolistic power by forming cartels enforced by criminal organizations in the adjudication of public contracts,2 and
by De Feo and De Luca (2017), who provide evidence on the support that Mafia can
give to local politicians in exchange for economic benefits. In addition, indirect
channels can also be at work. This occurs if the expected returns from organized
crime, including possible punishment and opportunity costs, are higher than the
expected return of the individuals’ legal alternatives. Our conjecture is that there
might be more gains from involvement in organized crime in the presence of
inequality as the latter is likely to lead to corruption and erosion of the rule of law
(Jong-Sung and Khagram 2005; Sunde et al. 2008).
In addition, we expect that low socio-economic mobility favors organized crime
development. This fact, as noted, is suggested by the empirical analyses of Acciari
et al. (2017), and Guell et al. (2017), although in these articles this issue is not considered. Our hypothesis is motivated by the ideas of memberships theory of inequality put forward by Durlauf (1996, 1999, 2006). Cross-sectional inequality leads to
segregation that generates gaps in the life-time outcomes of children that in turn
generate intergenerational immobility. Durlauf shows how the co-existence of social
interactions and residential segregation can lead to intergenerational persistence.
The idea is that in the presence of complementarity, parents (e.g., role models) will
make choices about the locations where their children grow up to affect their social
environment. This leads to neighborhood stratification by income and incentives for
segregation. Segregation in turn creates great gaps in the capabilities and outcomes
of children as they grow up to become adults. The higher the degree of segregation,
the larger are the gaps in human capital between children from rich and poor neighborhoods. This creates the Great Gatsby Curve (Durlauf and Seshadri 2018). More
precisely, assuming local public finance of education, rich families will sort themselves into rich and low-crime neighborhoods with high-quality schools and services. Poor families will sort themselves into high-crime neighborhoods with poor
schools and services. This implies that a child who grows up in a disadvantaged
neighborhood will receive poor education and it more likely to be exposed to a host
of negative social interaction effects such as the lack of exemplar role models and
negative peer effects that can influence educational and health outcomes or lead to
delinquent behavior. In contrast, favorite outcomes are expected for the human capital and skill formation of the children raised in rich neighborhoods. These gaps are
then transmitted from generation to generation giving rise to intergenerational
persistence.
Our hypothesis is that in such neighborhoods Mafiosi can represent relevant role
models, so that members of young generations have an incentive to fill the ranks of
See also Gambetta (1993) for an analysis of the Sicilian Mafia as a provider of private protection
and Lavezzi (2014) for a discussion of the concept of “demand for Mafia”.
2
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
207
criminal organizations. On the other hand, rich families tend to cluster to exploit
positive spillovers from other rich families. In connection with the argument we
made to motivate our hypothesis on the relationship between inequality and development of organized crime, we argue that in dynamic setting, offspring of rich families can have an incentive to exploit the same channels that members of previous
generations had to increase income and accumulate wealth, including the utilization
of the services provided by organized crime. Although our interest is on the causal
link going from social mobility to organized crime development, important feedback mechanisms can be at work, so that at this stage we cannot make strong claims
of causality and just look for robust correlations between measures of socio-­
economic mobility and of organized crime.3
Indeed, the few existing works investigating the relationship between organized
crime and social mobility study the causal link going from the former to the latter.
Coniglio et al. (2010) and Caglayan et al. (2017) in particular show that the presence
of organized crime within the territory interferes with human capital accumulation.
In particular, Coniglio et al. (2010) argue that the presence of organized crime in
Southern Italy, especially in the region of Calabria, hinders human capital accumulation by reducing incentives to invest in education and by increasing migration
outflows. Caglayan et al. (2017) instead provides empirical evidence of less accumulation of human capital in provinces of Northern Italy where the presence of
organized crime is widespread.4
In our empirical analysis we construct different measures of organized crime,
which are based on Calderoni (2011) and on dynamic factor methods (Moench et al.
2013; Bańbura and Modugno 2014). Furthermore, we use different inequality indices. Our main finding is that higher inequality leads to higher organized crime
development. The results are robust for different organized crime measures and
inequality indices. For all measures the ratio of 90th quintile over the 10th in the
distribution is a strong predictor of organized crime, even if we control for other
covariates that capture economic development, education, etc. We also find that
consumption inequality performs better that income inequality as the relevant
inequality measure. Finally, we conduct a provincial level analysis to study the
effect of socio-economic mobility on organized crime, using the Calderoni-based
indices. We consider three alternative measures of social mobility. We find that
lower socio-economic mobility displays a robust association with organized crime
development.
Work is in progress to analyze this issue in the city of Palermo.
Other works investigate the effect of other types of crime on social mobility. For example,
Savolainen et al. (2014), using birth cohort data from Finland, examine the relationship between
intergenerational educational mobility and criminal activity noting that rates of criminal offending
is strongly related to family background. Differently, Sharkey and Torrats-Espinosa (2017), using
longitudinal data on US, find evidence that a decline on violent crime in a county increases the
upward economic mobility in adults who had experienced this drop during the late adolescence.
Moreover, they show that a decline in the violent crime rate reduces the high school dropouts at the
county level.
3
4
208
M. Battisti et al.
This chapter is organized as follows. Section “Data” contains the description of
the dataset; Section “Methodology” describes the methodology employed for the
empirical analysis; Section “Results” presents the results; Section “Discussion”
contains a discussion of our results; Section “Conclusions” contains some concluding remarks.
Data
Our dataset on organized crime and inequality takes the form of an unbalanced
5-year period panel for 20 Italian regions for the period 1985–2014. A detailed
description of the data and our sources is in Table 1. While we construct the organized crime variables at the annual frequency we opt to use 5-period averages to
reduce measurement error bias and ease the problem of missing observations in
several explanatory variables.
Data on income inequality are obtained from the Survey on Household Income
and Wealth (SHIW) provided by the Bank of Italy, which comprises about 8000
households (20,000 individuals) per wave. This survey includes information on personal income, household consumption and wealth, education and occupation. Data
on socio-economic mobility are from Acciari et al. (2017), and come from Italian
tax data at provincial level.
Measures of Organized Crime
To measure organized crime, we follow the current literature and utilize data on
crimes committed by members of criminal organizations. Data on crimes come
from Istat and the SDI database, managed by the Italian Ministry of Interiors.5 Our
analysis distinguishes between two types of crime measures. Type I includes direct
measures, i.e. crimes that unambiguously are related to organized crime (2 crimes):
Homicide by Mafia and Mafia Type Association. Type II includes indirect measures,
i.e. crimes that are potentially associated to organized crime (14 crimes): Criminal
Association, Bribery, Corruption for an Act Against Official Duties, Drugs,
Extortion, Money Laundering, Prostitution, Smuggling, Threats, Usury, Corruption
in Public Acts, Instigation to Corrupt, Judicial Corruption, Kidnapping for Extortion
Purposes.6
Our organized crime measures are based on the joint consideration of type I and
II crimes. We choose to measure organized crime in this way as the relevance of
crimes of Type II is remarked in several reports on the activities of Italian mafias
We are grateful to Magg. Domenico Martinelli and to Claudia Di Persio for invaluable help, and
to the Department on “Analisi Criminale della Direzione Centrale della Polizia Criminale” at the
Italian Ministry of Interiors for releasing the data.
6
In our empirical analysis crime numbers are normalized by population.
5
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
209
Table 1 Variables’ definition and sources
Variable
Absolute Upward
mobility – Q1Q5
Absolute Upward
mobility – Expected rank
Bribery
Compensation of
employees
Corruption for an act
against Official Duties
Source
Acciari et al. (2017)
Acciari et al. (2017)
Concussione – Offences reported by the police to the prosecution
service (2003–2015) – SDI (acronym for System of Investigation)
Cambridge Econometrics (€2005 m)
Corruzione per un atto contrario ai doveri d’ufficio – Offences
reported by the police to the prosecution service (2003–2015) – SDI
(acronym for System of Investigation)
Corruption in public acts Corruzione per un atto d’ufficio – Offences reported by the police to
the prosecution service (2003–2015) – SDI (acronym for System of
Investigation)
Criminal_association
Associazione a delinquere – reported by the police forces to the
judicial authority (1985–2015) – ISTAT
Drugs
Produzione, commercio, ecc. di stupefacenti – reported by the police
forces to the judicial authority (1985–2015) – ISTAT
Economic Activity
Percentage of total population, both employed and unemployed, that
rate – total
constitutes the manpower supply of the labor market – Eurostat
Extortions
Estorsioni – reported by the police forces to the judicial authority
(1985–2015) – ISTAT
Gross fixed capital
Cambridge Econometrics (€2005 m)
formation
Homicide_Mafia
omicidi per motivi di mafia, camorra o ‘ndrangheta – Murder
committed by Mafia reported by the police forces to the judicial
authority (1985–2015) – ISTAT
Homicide_terrorism
Omicidi a scopo terroristico (art. 28- C.P.) – Murders committed by
Terrorists reported by the police forces to the judicial authority
(1985–2015) – ISTAT
Inequality – Net Income Gini and other inequality measures for Net Income – Bank of Italy
(SHIW)
Inequality – Total
Gini and other inequality measures for total consumption – Bank of
consumption
Italy (SHIW)
Inequality – Total
Gini index for total consumption (provincial level) – ISTAT –
consumption
Household budget survey – Processing carried out at the Laboratory
for Analysis of ELEmentary data (Adele Palermo – ISTAT)
Instigation to corrupt – Offences reported by the police to the
Instigation to corrupt
prosecution service (2003–2015) – SDI (acronym for System of
(Istigazione alla
Investigation)
corruzione)
Judicial corruption
Corruzione in atti giudiziari – Judicial corruption – Offences reported
by the police to the prosecution service (2003–2015) – SDI (acronym
for System of Investigation)
Kidnappings for
Sequestro di persona a scopo estorsivo – kidnappings for extortion
extortion purpose
purpose reported by the police forces to the judicial authority
(1985–2003) – ISTAT
(continued)
210
M. Battisti et al.
Table 1 (continued)
Variable
Source
Kidnappings for terrorist Sequestro di persona a scopo terroristico – kidnappings for terrorist
purposes
purposes reported by the police forces to the judicial authority
(1985–2003) – ISTAT
Long-term uneployment Eurostat
rate
Mafia type Association
Associazione a delinquere di stampo mafioso – Mafia criminal
association reported by the police forces to the judicial authority
(1985–2015) – ISTAT
Money laundering
Riciclaggio di denaro – Money laundering – Offences reported by the
police to the prosecution service (2003–2015) – SDI (acronym for
System of Investigation)
Participation rate in
Participation rate in education and training (last 4 weeks). Population
education and training
aged 25–64 – Eurostat
Prostitution
Prostituzione – Prostitution reported by the police forces to the
judicial authority (1985–2015) – ISTAT
Relative mobility
Acciari et al. (2017)
Robberies
Rapina – Robberies reported by the police forces to the judicial
authority (1985–2015) – ISTAT
Secondary Education
Percentage of people aged 25–64 with upper secondary and
level
post-secondary non-tertiary education attainment (levels 3 and
4) – Eurostat
Smuggling
Contrabbando – Smuggling reported by the police forces to the
judicial authority (1985–2015) – ISTAT
Theft
Furto – Theft reported by the police forces to the judicial authority
(1985–2015) – ISTAT
Threats
Minacce – Threats – Offences reported by the police to the
prosecution service (2003–2015) – SDI (acronym for System of
Investigation)
Total hours worked
Cambridge Econometrics (millions)
Usury
Usura – Usury – Offences reported by the police to the prosecution
service (2003–2015) – SDI (acronym for System of Investigation)
(see, e.g. Dia 2016), as well as in the literature (see e.g. Riccardi et al. 2016, and
Fioroni et al. 2017). This approach distinguishes us from, e.g., Calderoni (2011)
who focuses on type I crimes only.7 In addition, from a statistical point of view, we
see organized crime as a latent variable, measured with error. In this perspective, the
higher the number of crimes we consider to measure organized crime, the lower the
measurement error.
Our organized crime variables are based on two alternative approaches using 16
crime variables. The first one is based on Calderoni (2011) and the second approach
employs novel dynamic factor methods proposed by Moench et al. (2013) and
Banbura and Modugno (2014).
Crime data in Calderoni (2011) are integrated by data on other direct measures of Mafia activities:
the number of city councils dissolved for Mafia infiltration, and a measure of assets confiscated to
Mafia clans. See the same article for details on other methods to measure organized crime found in
the literature.
7
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
211
Table 2 Summary statistics of 5-year averages of crime indices
Calderoni Mean Index (Type
I,II)
Calderoni Rank Index (Type
I,II)
Obs. Mean Min
Max
sd
p25
p50
p75
120 30,378 9150 61,367 11,533 21,623 28,965 37,839
120
52,500 14,813 79,750 13,018 44,125 54,063 60,594
Following Calderoni (2011), we employ two alternative procedures to calculate
the Mafia Index. In the first one, we first normalize each variable to take values
between 0 and 100 with 100 to denote the highest value of Mafia presence. Then, we
compute the Calderoni Mean Index as the average score of all variables. One problem with the average score is that it may overestimate the presence of the mafia in
the Southern regions and underestimate it in the other regions. This is because organized crime is prevalent in the areas of southern Italy. To overcome this problem we
also construct an index based on region’s rank. For each variable, we rank all the
Italian regions in decreasing order. Then we attribute the score of 100 to the region
with the highest rank and proportionally lower scores to the other provinces, according to their rank. Then, we compute the Calderoni Rank Index as the average score
of all variables (see Table 2 for summary statistics).
In addition, we employ the dynamic factor model method to extract a latent factor for organized crime at the regional level. This methodology has not been applied
to such a purpose before. Let xt = (x1t, x2t, …, xnt)′, t = 1, . . , T denote a stationary
n-dimensional vector of crime series standardized to mean 0 and unit variance. We
assume that the factor model takes the following form:
x t = Λft + e t
(1)
ft = Aft −1 + ηt
(2)
where ft is a r × 1 vector of latent common factors of organized crime
and et = (e1t, e2t, …, ent)′, is the idiosyncratic component, uncorrelated with ft at all
leads and lags.8 The errors ηt are assumed to be innovations to the factor. The common component is given by the n × r matrix Λ, which contains factor loadings.
Following Banbura and Modugno (2014) we estimate the dynamic factor model
using a modified Expectation Maximisation (EM) algorithm. The idea of the algorithm is to write the likelihood as if the data were complete and to iterate between
two steps: in the Expectation step the missing data are filled in the likelihood, while
in the Maximisation step this expectation is re-optimized. This modification is
important in our analysis as it allows us to consider a larger number of time-series
that organized crime regardless of whether they have missing observations or not.
In addition to the latent factor at the regional level we extract factors at the macroarea (Centre, North and South) and Italy level for descriptive purposes. In doing so,
we employ two complementary factor analyses. First, we apply the aforementioned
methodology separately at different levels of aggregation. Figures 1 and 2 provide
8
The errors were not allowed to be serially correlated, but this assumption can be relaxed later.
212
M. Battisti et al.
Fig. 1 Factor for Organized crime based on Banbura and Modugno (2014) – Italy-wide Factor for
Organized Crime
Fig. 2 Factor for Organized crime based on Banbura and Modugno (2014) – Center, North, South
Factor levels
the t-plots of the factors for results for Italy-wide and the macro-area factors, respectively. It can be observed that there are some large spikes at the beginning of the
period, indicating a higher intensity of the phenomenon, while fluctuations are much
smaller in the subsequent years.
Table 3 presents the estimates the loadings of the factors and Table 4 the square
loadings (as % of total) to illustrate the relative importance of the different crimes.9
Table 4 suggests that some of the indirect measures of organized crime may have a
high capacity to capture the phenomenon, and that the effect is different across the
different Italian macro regions. For example, different corruption crimes stand out
(see, e.g. Fioroni et al. 2017, on corruption and organized crime), as well as money
laundering in the South (see Barone and Masciandaro 2011, for an analysis of organized crime and money laundering).
The factor loadings (estimated as stationary, non time-varying, values) can be interpreted as
weights telling how much the factor intensity depends (loads) on each crime over the period of
interest.
9
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
213
Table 3 Factor loadings computed following Banbura and Modugno (2014)
Bribery
Corruption for an action against official duties
Criminal association
Drugs
Extortions
Homicide by Mafia
Mafia type Association
Money laundering
Prostitution
Smuggling
Threats
Usury
Corruption in public acts
Instigation to corrupt
Judicial corruption
Kidnapping for extortion purpose
Italy
−3.966
−0.007
0.765
0.032
0.242
−0.168
−1.004
−2.453
−0.081
−0.012
−2.535
0.687
2.102
−0.397
9.423
−0.225
North
4.294
−3.089
−0.406
−0.007
−0.098
−0.146
0.741
−0.377
0.009
0.013
2.068
0.503
−1.473
−2.336
−5.595
0.068
Center
−31,185
−7.915
−3.635
−0.942
−0.516
−0.732
5.060
28.086
0.438
−0.088
−9.125
−21.516
−55.867
19.441
−26.330
−0.757
South
−0.257
−4.077
0.510
0.019
0.219
−0.163
−0.737
−6.729
−0.086
−0.006
−1.418
−0.850
1.696
−1.327
1.178
−0.220
Table 4 Square of factor loadings computed following Banbura and Modugno (2014), as
percentage of the total
Italy (%)
Bribery
12.71
Corruption for an action against official duties 0.00
Criminal association
0.47
Drugs
0.00
Extortions
0.05
Homicide by Mafia
0.02
Mafia type Association
0.81
Money laundering
4.86
Prostitution
0.01
Smuggling
0.00
Threats
5.19
Usury
0.38
Corruption in public acts
3.57
Instigation to corrupt
0.13
Judicial corruption
71.75
Kidnapping for extortion purpose
0.04
North (%) Centro (%) South (%)
25.49
14,73
0,09
13.20
0.95
23.20
0.23
0.20
0.36
0.00
0.01
0.00
0.01
0.00
0.07
0.03
0.01
0.04
0.76
0.39
0.76
0.20
11.94
63.19
0.00
0.00
0.01
0.00
0.00
0.00
5.91
1.26
2.80
0.35
7.01
1.01
3.00
47.26
4.01
7.55
5.72
2.46
43.27
10.50
1.94
0.01
0.01
0.07
Second, we employ the hierarchical dynamic factor models proposed by Moench,
Ng, and Potter (2013) to account for the correlations between the shocks at the different
levels of aggregation. In particular, we employ this multilevel factor model using common (Italy level – global), block-specific (macro regions), and sub-block-­specific
(regions) to capture the within- and between-block variations in the measurement sys-
214
M. Battisti et al.
Fig. 3 Factor for Organized crime based on Moench, Ng, and Potter (2013) – Italy
Fig. 4 Factor for Organized crime based on Moench, Ng, and Potter (2013) – Macro Regions
tem of organized crime. Given that this method does not allow for missing observations, we only use a set of five variables (Criminal Association, Drugs, Mafia type
Association, Extortions and Prostitution). As it is illustrated by Figs. 3 and 4, the common factor exhibits a similar pattern to the one we obtained using the first method in
Figs. 1 and 2 although it is only based on a much smaller set of variables. Table 5
presents the variance decomposition analysis to document the importance of the variation of aggregate (Italy-wide), block-specific (Macro), and subblock-specific (Regional)
components as well as idiosyncratic noise relative to the total variation in the data.
The variations due to idiosyncratic shocks dominate other variations in all cases.
Overall we can see that regarding the other variations the sub-block specific variations are larger. Notice in particular that in Southern regions (where organized crime
is more widespread) the regional component is relatively high.
215
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
Table 5 Share F, Share G, Share H, Share Z denote the average variance share across all variables
in the block due to aggregate, block-level, subblock-level and idiosyncratic shocks respective
Center Lazio
Marche
Toscana
Umbria
North Emilia
Romagna
Friuli Venezia
Giulia
Liguria
Lombardia
Piemonte
Trentino Alto
Adige
Valle D’Aosta
Veneto
South Abruzzo
Basilicata
Calabria
Campania
Molise
Puglia
Sardegna
Sicilia
Share F
Aggregate
component
(Italy)
0.10
0.08
0.10
0.19
0.17
Share G
Block –specific
component
(Macro-region)
0.09
0.07
0.09
0.16
0.07
Share H
Sub-block specific
component
(Region)
0.10
0.13
0.09
0.10
0.07
Share Z
Idiosyncratic
noise
0.71
0.72
0.72
0.55
0.69
0.10
0.04
0.17
0.69
0.21
0.20
0.20
0.17
0.09
0.09
0.08
0.07
0.10
0.03
0.06
0.06
0.61
0.68
0.66
0.70
0.15
0.19
0.02
0.00
0.03
0.01
0.00
0.01
0.01
0.01
0.06
0.08
0.13
0.02
0.14
0.07
0.02
0.03
0.04
0.06
0.09
0.06
0.13
0.25
0.15
0.22
0.22
0.22
0.23
0.21
0.70
0.66
0.71
0.72
0.69
0.70
0.75
0.75
0.72
0.72
Inequality and Socio-Economic Mobility
We employ six different inequality metrics that capture different aspects of the distribution: (i) the Gini index, (ii) the Atkinson inequality index,10 and four percentile
ratios: (iii) P90/P10, (iv) P75/P25, (v) P90/P50, and (vi) P75/P50. The Gini and
Atkinson indices take value between 0 and 1 as opposed to the percentile ratios that
do not have this restriction. For each inequality metric, we construct 2 different
measures of inequality based on total consumption and household net income. For
inequality measures on consumption and income we consider all the observations in
which the heads of household are aged from 25 to 65. Table 6 presents summary
statistics of these inequality measures.
For socio-economic mobility, we use the measures computed by Acciari et al.
(2017), based on Italian tax data. In particular our measures of mobility measure
10
For definitions and details on the Gini and Atkinson index see e.g. Cowell (2011).
216
M. Battisti et al.
Table 6 Summary statistics of inequality indices
Variable
Obs.
A. Net income – Households
Gini Index
120
Atkinson inequality index
120
P90/P10
120
P75/P25
120
P90/P50
120
P75/P50
120
B. Total Consumption – Households
Gini Index
118
Atkinson inequality index
118
P90/P10
118
P75/P25
118
P90/P50
118
P75/P50
118
Mean
sd
Min
Max
0.287
0.265
3.733
1.92
1.896
1.382
0.037
0.066
0.814
0.192
0.24
0.087
0.165
0.095
2.111
1.567
1.405
1.224
0.396
0.43
7.087
2.568
2.677
1.628
0.251
0.186
3.084
1.765
1.799
1.335
0.029
0.036
0.419
0.125
0.146
0.053
0.138
0.062
1.961
1.35
1.366
1.152
0.327
0.296
4.82
2.107
2.26
1.432
immobility between parents’ income in 1998 and children’s income in 2012 at provincial level. The three indices we consider are: (i) Relative mobility: this is the
slope of a rank-rank regression between child ranks and parent ranks and measure
the difference in outcomes between children from top vs. bottom income families
within province (Chetty et al. 2014). A high value indicates low mobility; (ii)
Absolute mobility – expected rank: this index measures the expected rank of children from families at the bottom 25% of the national parent income distribution
(Chetty et al. 2014). A high value indicates high mobility; (iii) Absolute mobility –
Q1toQ5 measures the probability of rising from the bottom quintile to the top quintile of the income distribution (Corak and Heisz 1999; Hertz 2006). A high value
indicates high mobility.
Other Covariates
We consider as covariates in our analysis indicators of the economic conditions and
of human capital, under the assumption that bad economic conditions and low levels
of human capital can positively affect organized crime development. In particular,
we use secondary education level, the economic activity rate, the growth rate of
compensation of employee, the growth rate of total hours worked, the participation
in education and training, the long-term unemployment rate, the growth rate of
gross fixed capital formation (see Tables 1 and 7 for the definitions and for summary
statistics).
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
217
Table 7 Summary statistics for the other covariates
Variable
Employment
Gross fixed capital formation
Compensation of employees
Primary Education level
Secondary level
Total active population
Total hours worked
Economic activity rate
Participation rate in education and training
Long-term unemployment rate
Employment
Obs.
120
120
120
58
58
120
120
76
58
71
120
Mean
sd
0.003
0.010
0.006
0.04
0.007
0.019
47.978
7.477
38.886
4.909
0.005
0.007
0
0.011
53.797
5.57
6.06069 9.47
5.991
17.800
0.003
0.010
Min
−0.024
−0.119
−0.065
32.5
29.56
−0.02
−0.03
43.5
1.187057
4.398
−0.024
Max
0.028
0.078
0.05
61.5
48.11
0.018
0.03
63.75
3.76
0.780
0.028
Methodology
The focus of this paper is to investigate how changes in organized crime (OC) are
related to inequality (INEQ) and social mobility (SM).
Our benchmark model for inequality takes the form of a dynamic panel model:
OCit = ρ OCit −1 + β I INEQit + +γ ′Zit + ξ t + vi + uit ,
(3)
where vi is the fixed effect, ξt is the time effect, and uit is the idiosyncratic error term.
Zit includes a constant and other covariates. Following Blundell and Bond (1998),
we estimate this model using system 2-step GMM up to 4 lags as instrumental variables and robust standard errors.11 Given that the impact of inequality on organized
crime development can be not contemporaneous, we also tried specifications with a
lagged value of INEQ.
In the case of measurement of organized crime by the estimated dynamic factor,
we had to consider differences in this variable to make the series stationary. This
implied considering also INEQ in differences. In this case the estimated dynamic
panel takes the form:
∆OCit = ρ∆OCit −1 + β I ∆ INEQit + γ ′∆ Zit + ξ t + vi + uit
(4)
As for socio-economic mobility, given that our data are only available as a cross-­
section we estimate a simple cross-sectional regression of the form:
OCi = β M SM i + γ ′Zi + ui
The next section contains the results of our econometric analysis.
11
We implement this estimation using the Stata package xtabond2 by Roodman (2009).
(5)
218
M. Battisti et al.
Results
This section presents the results of our econometric analysis. In particular, Section
“Organized Crime and Inequality” discusses the results on inequality and organized
crime, while Section “Organized Crime and Socio-Economic Mobility” those on
socio-economic mobility and OC.
Organized Crime and Inequality
Tables 8 and 9 contain the results of regressions of organized crime on INEQ for
both indices based on Calderoni (2011), i.e. Calderoni Rank Index and Calderoni
Mean Index, considering respectively indices of income inequality and consumption inequality.12 The results show that the coefficients of the effect of inequality on
organized crime have positive and significant coefficients in particular when the
Calderoni Mean Index is utilized in the regressions. Under the null of joint validity
of instruments, the Hansen’s test of over-identification provides us enough evidence
about the validity of our instruments, in all models (at least at 5% significance level).
We then estimated specifications in which we added lagged values of inequality
indices. Table 10 contains the results of specifications for consumption inequality, as
the effects of income inequality resulted scarcely significant.13 Results in Table 10
shows that, with the Calderoni Mean Index, all six metrics of inequality measures have
positive and significant coefficients: in particular, the contemporaneous values of P90/
P10, P90/P50 and P75/P50 and the lagged values of the Gini and Atkinson Indices and
P75/P25. In the case of the Calderoni Rank Index, the coefficients of the lagged values
of the Atkinson index, P90/P10, P75/P25, P90/P50 are positive and significant.
Given the results presented Tables 8, 9 and 10 we kept as our preferred metric of
inequality the P90/P10 index and its lagged value.14 Table 11 contains the results of
regressions in which we added other covariates, related to potential economic deter-
Regressions are run on data from 19 out of 20 Italian regions. The region of Val d’Aosta is
dropped for lack of data.
13
In particular, results for income inequality return a significant coefficient only with the Calderoni
Mean Index for the lagged values of the Atkinson index (at 5%) and of P90/P10 (at 1%).
14
In particular, we chose P90/P10 as our preferred (consumption) inequality measure for the following reasons. P75/P50 in Tables 7 and 8 is significant 4 times out of 4, while P90/P10 is significant 3 times out of four. However, in Table 9, P75/P50 is scarcely significant. Moreover, in Table 9
we see that P90/P10 is significant in 2 cases out of 4 (considering both the coefficients for contemporaneous and lagged inequality indices), as other inequality measures, but it is the only one with
a highly significant coefficient in at least one case. This result is likely to depend on the fact that
the relationship between the extreme quintiles is more representative of the dynamics of inequality
over time than measures based on quintiles that mostly capture the dynamics of the share of income
of the middle class (see e.g. Garbinti et al. 2018)
12
period6
period5
period4
period3
P75/P50
P90/P50
P75/P25
2.47∗∗
(0.90)
(3)
0.72∗∗∗
(0.06)
8.19
(7.37)
(4)
0.74∗∗∗
(0.20)
8.44∗
(4.20)
(5)
0.72∗∗∗
(0.09)
(6)
0.69∗∗∗
(0.10)
17.78∗∗
(8.13)
6.78∗∗∗ 7.56∗∗∗ 6.36∗∗∗ 7.11∗∗∗ 6.45∗∗∗ 7.51∗∗∗
(1.67)
(1.82)
(1.46)
(1.55)
(1.38)
(1.04)
11.55∗∗∗ 11.48∗∗∗ 10.88∗∗∗ 11.32∗∗∗ 10.07∗∗∗ 11.18∗∗∗
(1.57)
(1.95)
(1.56)
(1.79)
(1.73)
(1.57)
4.85∗
4.18
3.85∗∗
3.86∗
4.27∗
4.64∗∗∗
(2.67)
(2.64)
(1.74)
(2.20)
(2.18)
(1.47)
5.65∗∗∗ 5.44∗∗
4.43∗∗
5.05∗∗∗ 4.93∗∗∗ 6.21∗∗∗
Dependent var.
Mafia rank index
Regressors
(1)
(2)
Lagged Mafia Index 0.60∗∗
0.72∗∗∗
(0.27)
(0.17)
Gini
45.96
(61.63)
Atkinson
2.10
(19.49)
P90/P10
−0.68
(0.98)
0.42
(0.87)
0.55
(2.66)
−0.11
−0.98
(1.34)
0.08
(0.94)
−0.02
(2.54)
−1.87
−1.08
(1.06)
0.77
(1.01)
0.96
(2.73)
−1.01
−0.06
(1.17)
2.07∗
(1.01)
1.22
(2.73)
−0.04
Mafia mean index
(7)
(8)
(9)
(10)
0.87∗∗∗ 0.96∗∗∗ 0.88∗∗∗ 0.95∗∗∗
(0.08)
(0.08)
(0.06)
(0.07)
76.51
(56.72)
33.16
(25.84)
3.53∗∗∗
(1.04)
19.70∗∗
(8.08)
−0.49
(1.05)
0.64
(1.08)
2.08
(2.69)
1.12
19.80∗∗
(8.50)
(continued)
48.77∗∗∗
(16.61)
0.68
(1.18)
1.37
(1.05)
1.41
(2.73)
1.06
(11)
(12)
0.95∗∗∗ 0.91∗∗∗
(0.09)
(0.09)
Table 8 This tables presents dynamic panel regressions of Mafia Rank Index and Mafia Mean Index based on Calderoni (2011) on alternative consumption
inequality indices. All results are based on the baseline sample of 5-year averages. All models control for fixed effects and time period effects. Estimation is
based on system GMM (Blundell and Bond (1998) using instruments up to the 4th lag. Each cell reports the coefficient estimates and robust standard errors in
the parenthesis. ***, **, and * denote significance of the regression coefficient at 1%, 5%, and 10%
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
219
Mafia rank index
(1)
(2)
(1.55)
(2.04)
−6.55
2.46
(11.44)
(4.88)
95
95
19
19
0.818
0.730
(3)
(1.57)
−5.71
(3.59)
95
19
0.925
(4)
(1.37)
−13.01
(9.30)
95
19
0.761
(5)
(1.51)
−11.58
(7.07)
95
19
0.986
(6)
(1.39)
−20.20∗
(9.81)
95
19
0.749
Mafia mean index
(7)
(8)
(1.45)
(2.31)
−15.51 −6.46
(18.42) (10.30)
95
95
19
19
0.540
0.387
(9)
(1.39)
−7.02
(6.28)
95
19
0.446
(10)
(1.28)
−35.74∗∗
(16.38)
95
19
0.555
(11)
(1.37)
−35.22∗
(17.73)
95
19
0.554
(12)
(1.38)
−63.34∗∗
(25.81)
95
19
0.511
All results are based on the baseline sample of 5-year averages. All models control for fixed effects and time period effects. Estimation is based on system
GMM (Blundell and Bond (1998) using instruments up to the 4th lag. Each cell reports the coefficient estimates and robust standard errors in the parenthesis.
∗∗∗, ∗∗, and ∗ denote significance of the regression coefficient at 1%, 5%, and 10%
Observations
Number of Regions
Hansen test p-value
Constant
Dependent var.
Regressors
Table 8 (continued)
220
M. Battisti et al.
period6
period5
period4
period3
P75/P50
P90/P50
P75/P25
5.65
(5.74)
(3)
0.63∗∗∗
(0.16)
10.26
(14.35)
(4)
0.75∗∗∗
(0.12)
2.72
(6.86)
(5)
0.73∗∗∗
(0.10)
(6)
0.74∗∗∗
(0.13)
53.59∗∗
(24.95)
8.30∗∗∗ 6.66∗∗∗ 6.97∗∗∗ 8.28∗∗∗ 7.61∗∗∗ 6.09∗∗∗
(1.35)
(1.72)
(1.50)
(1.42)
(1.21)
(1.34)
11.94∗∗∗ 10.19∗∗∗ 10.99∗∗∗ 11.98∗∗∗ 11.41∗∗∗ 9.82∗∗∗
(2.25)
(2.63)
(1.97)
(2.66)
(2.38)
(1.75)
4.38∗
7.88∗∗∗ 5.51∗
4.67
4.24∗
3.33
(2.35)
(2.61)
(2.67)
(2.72)
(2.07)
(2.21)
5.61∗∗∗ 6.94∗∗∗ 6.00∗∗∗ 5.71∗∗
5.50∗∗∗ 5.36∗∗∗
Dependent var.
Mafia rank index
Regressors
(1)
(2)
Lagged Mafia Index 0.80∗∗∗ 0.46∗∗∗
(0.09)
(0.15)
Gini
41.41
(42.08)
Atkinson
82.54∗
(42.77)
P90/P10
(continued)
Mafia mean index
(7)
(8)
(9)
(10)
(11)
(12)
0.91∗∗∗ 0.99∗∗∗ 1.04∗∗∗ 0.98∗∗∗ 0.93∗∗ 0.94∗∗∗
(0.20)
(0.20)
(0.17)
(0.17)
(0.33) (0.18)
179.06∗∗
(71.75)
151.78∗∗
(61.62)
13.85∗∗∗
(4.51)
33.27
(19.90)
50.19
(30.88)
98.06∗∗
(37.02)
1.62
1.11
0.48
0.42
−0.39 −0.91
(2.20)
(2.47)
(2.12)
(1.72)
(2.10) (1.80)
1.51
1.72
−0.29
0.11
0.87
−0.53
(2.10)
(2.39)
(2.67)
(2.13)
(2.09) (1.86)
3.17
4.58
3.98
1.11
4.83
0.47
(4.70)
(4.27)
(4.71)
(4.51)
(4.64) (3.68)
2.57
1.44
0.22
0.57
3.64
1.81
Table 9 This tables presents dynamic panel regressions of Mafia Rank Index and Mafia Mean Index based on Calderoni (2011) on alternative inequality
indices. All results are based on the baseline sample of 5-year averages. All models control for fixed effects and time period effects. Estimation is based on
system GMM (Blundell and Bond (1998) using instruments up to the 4th lag. Each cell reports the coefficient estimates and robust standard errors in the
parenthesis. ***, **, and * denote significance of the regression coefficient at 1%, 5%, and 10%.
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
221
(3)
(1.64)
−11.31
(15.53)
95
19
0.979
(4)
(2.00)
−16.27
(25.23)
95
19
0.986
(5)
(1.76)
−1.94
(12.39)
95
19
0.968
(6)
(1.50)
−68.27∗∗
(32.19)
95
19
0.998
Mafia mean index
(7)
(8)
(2.41)
(2.78)
−42.68
−30.40
(25.42)
(19.12)
95
95
19
19
0.103
0.0729
(9)
(2.18)
−45.61∗∗
(19.59)
95
19
0.0866
(10)
(2.57)
−58.36
(40.67)
95
19
0.0571
(11)
(3.33)
−88.77
(71.94)
95
19
0.169
(12)
(1.78)
−128.52∗∗
(57.21)
95
19
0.232
All results are based on the baseline sample of 5-year averages. All models control for fixed effects and time period effects. Estimation is based on system
GMM (Blundell and Bond (1998) using instruments up to the 4th lag. Each cell reports the coefficient estimates and robust standard errors in the parenthesis.
∗∗∗, ∗∗, and ∗ denote significance of the regression coefficient at 1%, 5%, and 10%
Mafia rank index
(1)
(2)
(1.71)
(1.47)
Constant
−9.86
−5.05
(11.01)
(11.76)
Observations
95
95
Number of Regions 19
19
Hansen test p-value 0.983
0.187
Dependent var.
Regressors
Table 9 (continued)
222
M. Battisti et al.
P90/P50
Lagged P75/P25
P75/P25
Lagged P90/P10
Dependent var.
Mafia rank index
Regressors
(1)
(2)
Lagged Mafia Index 0.42∗∗
0.42∗∗
(0.17)
(0.15)
Gini
78.50
(72.39)
Lagged Gini
41.35
(31.44)
Atkinson
57.76
(60.21)
Lagged Atkinson
48.23∗
(26.09)
P90/P10
1.25
(5.22)
3.81∗
(2.13)
(3)
0.35
(0.25)
−3.80
(13.20)
20.19∗
(11.46)
0.32
(12.23)
(4)
(5)
0.76∗∗∗ 0.50
(0.08)
(0.31)
(6)
0.47
(0.29)
Mafia mean index
(7)
(8)
(9)
(10)
(11)
0.86∗∗∗ 0.85∗∗∗ 0.92∗∗∗ 0.84∗∗∗ 0.92∗∗∗
(0.08)
(0.09)
(0.16)
(0.09)
(0.08)
32.85
(46.96)
70.80∗∗
(33.06)
20.39
(41.88)
53.51∗
(29.15)
15.28∗∗∗
(4.30)
−2.84
(3.46)
3.53
(15.78)
14.98∗
(8.27)
21.79∗∗
(9.80)
(continued)
(12)
0.84∗∗∗
(0.07)
Table 10 This table presents dynamic panel regressions of the Mean Mafia Index, which is the Mafia Rank Index based on Calderoni (2011), on P90/P10 ratio
and other determinants. All results are based on the baseline sample of 5-year averages. All models control for fixed effects and time period effects. Each cell
reports the coefficient estimates and robust standard errors in the parenthesis. All regression models include fixed effects and time effects. Estimation is based
on system GMM (Blundell and Bond (1998) using instruments up to the 4th lag. ***, **, and * denote significance of the regression coefficient at 1%, 5%, and
10%.
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
223
(5)
13.74∗
(7.27)
(6)
−1.43
(1.16)
−0.97
(1.10)
−0.47
(3.08)
0.54
(1.87)
−18.61
(14.26)
95
19
0.966
−1.28
(1.24)
−0.85
(1.23)
−0.33
(3.07)
0.37
(1.76)
−5.43
(10.20)
95
19
0.961
Mafia mean index
(7)
(8)
1.05
(1.54)
1.03
(2.19)
4.16
(3.18)
1.11
(1.23)
−35.72∗∗
(15.23)
95
19
0.344
(9)
−0.89
(1.23)
−0.59
(1.05)
−0.57
(2.43)
0.20
(1.56)
−23.77
(25.63)
95
19
0.981
(10)
(12)
86.02∗
(41.14)
−7.93
(26.16)
0.06
−1.28
(1.08)
(1.37)
−0.06
−0.71
(1.45)
(1.81)
2.31
1.70
(2.79)
(3.07)
2.27
0.85
(1.82)
(1.61)
−39.73∗∗ −96.18∗∗
(14.84)
(37.50)
95
95
19
19
0.910
0.989
(11)
1.97
(8.22)
All results are based on the baseline sample of 5-year averages. All models control for fixed effects and time period effects. Estimation is based on system
GMM (Blundell and Bond (1998) using instruments up to the 4th lag. Each cell reports the coefficient estimates and robust standard errors in the parenthesis.
∗∗∗, ∗∗, and ∗ denote significance of the regression coefficient at 1%, 5%, and 10%
Observations
Number of Regions
Hansen test p-value
Constant
period6
period5
period4
period3
Lagged P75/P50
31.88
(90.24)
5.63
(43.36)
5.55∗∗
5.09∗∗∗ 5.00∗∗∗ 5.37∗∗ 6.00∗∗∗ 5.48∗∗
(2.11)
(1.45)
(1.57)
(2.19)
(1.44)
(2.08)
10.05∗∗∗ 9.98∗∗∗ 10.75∗∗∗ 8.70∗∗∗ 9.72∗∗∗ 10.42∗∗∗
(1.97)
(1.83)
(1.49)
(1.96)
(2.64)
(2.16)
7.22∗∗
7.15∗∗ 6.45∗∗
1.48
4.12
5.84∗∗
(3.10)
(2.87)
(2.97)
(2.24)
(4.51)
(2.42)
7.85∗∗∗ 7.37∗∗∗ 7.49∗∗∗ 4.04∗∗ 6.91∗∗ 6.67∗∗∗
(1.87)
(1.71)
(1.77)
(1.74)
(3.01)
(1.69)
−17.60
−6.92
−1.11
−24.51 −15.19 −39.62
(19.88)
(13.03) (16.68)
(20.99) (30.31) (73.91)
95
95
95
95
95
95
19
19
19
19
19
19
0.360
0.405
0.518
0.675
0.549
0.399
(4)
P75/P50
(3)
Mafia rank index
(1)
(2)
Dependent var.
Regressors
Lagged P90/P50
Table 10 (continued)
224
M. Battisti et al.
Gross fixed
capital
formation
Long-­run
unemployment
Compensation
of employees
Secondary
Education level
Economic
activity rate
Lagged P90/
P10
(3.10)
(0.30)
−0.77∗∗
(1.98)
0.13
(0.05)
(2.07)
1.76
(1.38)
1.15
(0.08)
(0.15)
−0.20
(2.15)
1.52
(1.47)
1.16
(0.10)
(0.26)
−0.31
(2.26)
1.67
(1.94)
0.66
(12)
(0.05)
(13)
(1.88)
(14)
(1.19)
(30.03)
(37.95)
(0.14)
−0.24
(1.74)
0.75
−3.32
(0.55)
(0.06)
0.91
23.15
(0.72)
(0.21)
−0.19
(2.34)
0.89
0.30
(29.36)
(54.31)
(0.10)
0.48
−3.12∗∗∗
−32.20
(0.11)
−0.21∗
(1.84)
1.60
(1.52)
0.12
26.97
(0.44)
(0.25)
(0.11)
0.56
(0.45)
(0.20)
(11)
0.08
(0.21)
Mafia mean index
(9)
(10)
(15)
(16)
(0.10)
(0.25)
−0.35
(2.39)
1.47
(2.01)
0.85
(continued)
(0.15)
−0.21
(1.88)
0.46
(1.50)
1.43
(0.08)
0.84∗∗∗ 0.67∗∗∗ 0.65∗∗∗ 0.62∗∗∗ 0.70∗∗∗ 0.66∗∗∗ 0.67∗∗∗ 0.68∗∗∗ 0.65∗∗∗
(8)
−0.42
(0.29)
(0.20)
(1.06)
1.22
(2.59)
−2.12
(0.07)
0.85∗∗∗
(7)
(0.22)
(0.92)
1.12
(1.67)
−1.88
(0.07)
0.92∗∗∗
(6)
−0.71∗∗∗ −1.54∗∗∗ −0.75∗∗∗ −0.86∗∗∗
(2.28)
7.52∗∗∗
(3.22)
2.49
(0.08)
0.84∗∗∗
(5)
−0.79∗∗∗ −0.60∗
(0.98)
(1.78)
(1.18)
(1.11)
(1.86)
2.09∗∗
(3.31)
−0.07
(2.37)
(1.08)
(0.06)
−2.42
3.66∗∗∗ 2.06∗
(0.09)
−0.00
(0.06)
−1.99
0.63
(0.06)
P90/P10
0.85∗∗∗ 0.93∗∗∗
0.93∗∗∗ 0.92∗∗∗
(4)
Lagged Mafia
Index
(3)
Mafia rank index
(1)
(2)
Dependent var
Regressors
Table 11 This table presents dynamic panel regressions of the Mean Mafia Index, which is the Mafia Rank Index based on Calderoni (2011), on P90/P10 ratio and other
determinants
(24.61)
(2.16)
(1.21)
0.410
Hansen test
p-value
0.456
19
−2.42
(2.98)
(24.54)
47.61∗
(0.00)
0.00
(0.00)
0.167
19
55
(2.85)
0.943
19
72
(15.77)
43.43∗∗
(1.87)
(2.61)
(3.44)
0.902
19
70
(20.51)
0.987
19
72
(15.80)
0.997
19
72
(21.67)
67.91∗∗∗ 48.34∗∗∗ 57.95∗∗
0.0938
19
55
(0.00)
0.00
(29.45)
49.39
(1.61)
(2.31)
(2.21)
0.867
19
95
(5.10)
−4.39
(1.44)
0.628
19
72
(12.34)
11.79
(1.84)
4.23∗∗∗ 0.40
(1.85)
3.84∗
(2.12)
9.05∗∗∗ 4.30∗
(0.00)
0.309
19
55
(18.20)
17.63
(0.00)
0.00
(2.78)
−0.48
(3.99)
3.76
(1.90)
0.971
19
72
(9.83)
14.64
(2.02)
0.28
(1.93)
4.82∗∗
(3.48)
0.739
19
70
(14.42)
14.70
(2.16)
0.04
(2.56)
4.03
(2.60)
0.972
19
72
(11.94)
16.73
(2.02)
0.43
(2.72)
5.00∗
0.971
19
72
(11.35)
14.37
(3.11)
0.30
(2.57)
5.01∗
(2.27)
1.21
0.376
19
55
(18.27)
14.72
(0.00)
0.00
(2.85)
−0.17
(5.58)
5.38
(0.00)
0.00
(2.43)
1.53
(16)
(3.40)
0.42
(15)
All results are based on the baseline sample of 5-year averages. All models control for fixed effects and time period effects. Each cell reports the coefficient estimates and
robust standard errors in the parenthesis. All regression models include fixed effects and time effects. Estimation is based on system GMM (Blundell and Bond (1998) using
instruments up to the 4th lag. ∗∗∗, ∗∗, and ∗ denote significance of the regression coefficient at 1%, 5%, and 10%
19
Number of
Regions
(18.96)
72
(6.67)
95
0.00
(27.69)
(3.24)
47.21∗∗
(1.39)
(24.64)
59.12∗∗ 2.25
1.39
(2.34)
−11.42
2.15
(2.92)
1.22
(2.74)
0.82
1.71
(14)
1.08
0.00
(13)
−0.19
5.66∗∗∗ 0.93
(12)
(144.82)
(11)
(144.53)
Mafia mean index
(9)
(10)
2.52
(8)
−314.74∗∗
(7)
(26.40)
7.17∗∗∗
(1.63)
−2.76
(3.36)
−2.56
(6)
49.36∗
(2.24)
0.97
(2.24)
6.92∗∗∗
(5)
57.85∗∗
−0.37
−0.12
(2.61)
(0.00)
55.15∗∗ −2.31
(2.67)
−2.31
(1.23)
−1.84
−0.88
(4)
0.00
(3)
−1.64
−1.10
Mafia rank index
(1)
(2)
Observations
Constant
Period6
Period5
Period4
Period3
Part. in
education and
training
Total hours
worked
Dependent var
Regressors
Table 11 (continued)
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
227
minants of OC.15 Table 11 shows that the lagged value of P90/P10 has a positive and
significant coefficient in almost all the specifications with the Mafia Rank Index.
Table 12 presents the results where the measure of organized crime is based on
the innovative method we described in Section 2.1 based on the estimation of a
dynamic common factor.16 We can see that the coefficients of the indices of inequality (in this case appearing as differences) have positive and significant coefficients
with the exception of P75/P25 and P75/P50.
Table 13 contains the results of the regressions with additional covariates when
we consider the effect of consumption inequality on the factor index computed following Banbura and Modugno (2014). 17 We can observe that in almost all the specifications the effect of inequality (in differences) is positive and significant.
Organized Crime and Socio-Economic Mobility
In this section, we present the results of cross-section regressions estimating the
effect of indicators of socio-economic mobility on OC. The analysis is carried out
at provincial level. Table 14 contains the summary statistics. Table 15 contains simple regressions of our two organized crime measures (Calderoni Rank Index and
Calderoni Mean Index) on the three metrics of socio-economic mobility we considered in this report only and a constant, while Tables 16, 17 and 18 adds the covariates related to economic conditions and human capital levels to regressions with the
Calderoni Mean Index.18 The mobility indices’ coefficients have the expected signs
and are highly significant. The same result holds for Table 19, which reports the
results for the correlation of the relative mobility index on the Calderoni Rank Index.19
Discussion
The main findings of our study are twofold. First, it documents the linkages between
organized crime and inequality. We find that higher levels of inequality can lead to
higher organized crime development using the percentile ratio P90/P10 of conWe only report results for consumption inequality. Regressions on income inequality returned
positive and significant coefficients for the lagged value of P90/P10 in Models 1, 2, 4, 5 with the
Calderoni Rank Index.
16
We only report results for consumption inequality, as regressions on income inequality measures
did not return significant results.
17
We also estimated regressions using the Hierarchical dynamic factors, but the results were not
significant. This is likely due to the fact that we used only five variables to estimate them.
18
The set of covariates used for the regional and provincial analysis do not perfectly match because
of data availability. Model 7 in Tables 15, 16 and 17 contains a dummy for the Northern regions as
the latter proved to be highly significant in regressions on the mobility indices and macro-region
dummies.
19
The results for the other mobility indices are not significant.
15
Observations
Constant
Period6
Period5
Period4
Period3
ΔP75/P50
ΔP90/P50
ΔP75/P25
ΔP90/P10
ΔAtkinson
ΔGini
Dependent variable
Regressors
Lagged Mafia Index
0.01
(0.02)
0.01
(0.03)
0.03
(0.04)
0.01
(0.03)
−0.02
(0.02)
95
0.00
(0.02)
0.01
(0.04)
0.03
(0.04)
0.00
(0.02)
−0.01
(0.02)
95
−0.00
(0.03)
0.00
(0.04)
0.04
(0.06)
−0.01
(0.03)
−0.01
(0.03)
95
−0.03∗
(0.02)
−0.02
(0.03)
−0.02
(0.02)
−0.01
(0.02)
0.01
(0.01)
95
Hierarchical dynamic factor analysis
(1)
(2)
(3)
(4)
−0.43
−0.41
−0.37
−0.37∗
(0.25)
(0.25)
(0.32)
(0.19)
1.42∗∗
(0.64)
1.16∗∗
(0.53)
0.15∗
(0.08)
0.08
(0.16)
−0.00
(0.03)
0.01
(0.04)
0.04
(0.04)
0.02
(0.03)
−0.01
(0.02)
95
0.34∗∗
(0.14)
(5)
−0.32∗
(0.17)
−0.52
(0.40)
−0.04
(0.02)
−0.03
(0.03)
−0.05∗∗∗
(0.02)
−0.02
(0.02)
0.03∗
(0.01)
95
(6)
−0.26∗
(0.15)
0.06
(0.04)
0.04
(0.05)
0.09
(0.05)
0.03
(0.03)
−0.05
(0.04)
95
0.09
(0.06)
0.06
(0.07)
0.10
(0.06)
0.02
(0.03)
−0.06
(0.05)
95
EM algorithm
(7)
(8)
−0.09∗ −0.11
(0.05)
(0.08)
2.08∗
(1.02)
1.51∗
(0.77)
Table 12 The tables presents regressions of two latent factors on alternative inequality of consumption indices
0.07
(0.05)
0.04
(0.05)
0.10∗
(0.05)
0.01
(0.04)
−0.05
(0.04)
95
0.14∗∗
(0.07)
(9)
−0.16∗
(0.08)
0.02
(0.07)
0.01
(0.07)
0.05
(0.07)
−0.02
(0.04)
−0.01
(0.05)
95
0.23
(0.26)
(10)
−0.12
(0.08)
0.03
(0.04)
0.02
(0.06)
0.06
(0.04)
0.01
(0.03)
−0.03
(0.04)
95
0.24∗∗∗
(0.07)
(11)
−0.11
(0.07)
1.18
(1.19)
0.03
(0.08)
−0.01
(0.07)
0.06
(0.08)
0.00
(0.04)
−0.02
(0.06)
95
(12)
−0.12∗∗
(0.05)
228
M. Battisti et al.
Hierarchical dynamic factor analysis
(1)
(2)
(3)
(4)
19
19
19
19
0.280
0.297
0.574
0.445
(5)
19
0.553
(6)
19
0.307
EM algorithm
(7)
(8)
19
19
0.624
0.219
(9)
19
0.987
(10)
19
0.210
(11)
19
0.404
(12)
19
0.461
In Panel A, we have the results of a factor constructed using a hierachical dynamic factor methodology, while in Panel B we show the results of a dynamic
factor estimated by using a modified Expectation Maximisation (EM) algorithm. Given that the latent factors are constructed on transformed data to endure
stationarity (growth rates) we also transform the regressors into first differences, too. All results are based on the baseline sample of 5-year averages. Each cell
reports the coefficient estimates and robust standard errors in the parenthesis. All regressions control for fixed effects and time period effects. ∗∗∗, ∗∗, and ∗
denote significance of the regression coefficient at 1%, 5%, and 10%
Dependent variable
Regressors
Number of Regions
Hansen test p-value
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
229
Period6
Period5
Period4
Period3
ΔPart. in education and training
ΔTotal hours worked
ΔGross fixed capital formation
ΔLong-run unemployment
ΔCompensation of employees
ΔSecondary Education level
ΔEconomic activity rate
ΔP90/P10
Dependent variable
Regressors
Lagged Mafia Index
0.07
(0.05)
0.04
(0.05)
0.10∗
(0.05)
0.01
0.00
(0.00)
0.00
(0.00)
0.03
(0.05)
−0.06
Mafia factor index
(1)
(2)
−0.16∗
−0.48∗∗∗
(0.08)
(0.16)
0.14∗∗
0.07∗
(0.07)
(0.04)
−0.03
(0.03)
0.00
(0.00)
0.00
(0.00)
0.02
(0.05)
−0.02
−0.00
(0.02)
(3)
−0.48∗∗∗
(0.05)
0.02
(0.04)
0.06
(0.06)
0.01
(0.05)
0.07
(0.04)
−0.02
−0.08
(0.70)
(4)
−0.22∗∗
(0.08)
0.10∗∗
(0.04)
0.00
(0.00)
0.00
(0.00)
0.03
(0.04)
0.05
−0.02
(0.02)
(5)
−0.47∗∗∗
(0.11)
0.07∗∗
(0.03)
0.11∗∗
(0.05)
0.05
(0.04)
0.09∗
(0.04)
0.02
−0.52
(0.41)
(6)
−0.13
(0.08)
0.10∗
(0.05)
Table 13 This table presents dynamic panel regressions of the Mafia Factor Index on P90/P10 ratio and other determinants
0.05
(0.08)
0.04
(0.05)
0.08∗
(0.04)
0.00
0.03
(1.91)
(7)
−0.13∗
(0.07)
0.12∗∗
(0.05)
−0.02
(0.09)
0.00
(0.00)
0.00
(0.00)
0.00
(0.00)
−0.06
(8)
−0.43∗
(0.25)
0.05
(0.07)
230
M. Battisti et al.
Mafia factor index
(1)
(2)
(0.04)
(0.05)
−0.05
0.04
(0.04)
(0.03)
95
53
19
19
0.987
0.688
(3)
(0.05)
0.00
(0.00)
36
19
0.564
(4)
(0.03)
−0.02
(0.04)
95
19
1.000
(5)
(0.09)
−0.03
(0.05)
51
19
0.635
(6)
(0.02)
−0.07∗∗
(0.03)
95
19
1.000
(7)
(0.03)
−0.05
(0.04)
95
19
1.000
(8)
(0.11)
0.04
(0.13)
36
19
0.838
All results are based on the baseline sample of 5-year averages. All models control for fixed effects and time period effects. Given that the latent factors are
constructed on transformed data to endure stationarity (growth rates) we also transform the regressors into first differences, too. Each cell reports the coefficient
estimates and robust standard errors in the parenthesis. All regression models include fixed effects and time effects. Estimation is based on system GMM
(Blundell and Bond (1998) using instruments up to the 4th lag. ∗∗∗, ∗∗, and ∗ denote significance of the regression coefficient at 1%, 5%, and 10%
Observations
Number of Regions
Hansen test p-value
Constant
Dependent variable
Regressors
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
231
232
M. Battisti et al.
Table 14 Summary statistics on socio-economic mobility and OC measures at provincial level
Variable
A. Organized Crime Index
A.3 Calderoni Mean Index
(Type I,II)
A.4 Calderoni Rank Index (Type
I,II)
B. Social Mobility
B.1 Absolute Upward Mobility:
Expected Rank
B.2 Absolute Upward Mobility:
Q1Q5
B.3 Relative Mobility
N Mean
95
Min
2.003
Max
0.453 100
sd
10.163
p25
p50
0.820
0.970
p75
1.051
95 50.526 15.855 93.816 13.005 42.039 50.724 59.474
95
0.463
0.371 0.625
0.059
0.410
0.457
0.523
95
0.131
0.042 0.389
0.064
0.077
0.112
0.188
95
0.168
0.095 0.222
0.028
0.148
0.169
0.190
Table 15 The tables presents cross-sectional regressions of three alternative socio-economic
mobility measures
Dependent variable
Regressors
Absolute Upward
Mobility: Expected
Rank
Mafia mean index (Type I,II)
(7)
(8)
(9)
−60.35∗∗∗
Mafia rank index (Type I,II)
(10)
(11)
(12)
−53.19∗∗
(9.776)
(20.72)
−50.32∗∗∗
Absolute Upward
Mobility: Q1Q5
−39.19∗∗
(9.324)
Relative Mobility
Constant
Observations
R-squared
48.81∗∗∗
(4.778)
95
0.269
27.42∗∗∗
(1.471)
95
0.218
(18.56)
116.7∗∗∗
(23.62)
1.25
(3.901)
95
0.235
179.2∗∗∗
(42.37)
75.18∗∗∗ 55.65∗∗∗ 20.44∗∗∗
(9.849)
(2.842)
(7.253)
95
95
95
0.064
0.041
0.17
All results are based on the baseline sample of 5-year averages. Each cell reports the coefficient
estimates and robust standard errors in the parenthesis. ∗∗∗, ∗∗, and ∗ denote significance of the
regression coefficient at 1%, 5%, and 10%
sumption inequality. Second, we find that lower levels of socio-economic mobility
are associated with organized crime development.
An important issue is whether changes in welfare are better measured by consumption inequality rather than income inequality, especially when one examines
the role of inequality in organized crime. As argued by Attanasio and Pistaferri
(2016) consumption inequality is potentially a better measure of changes in welfare
than income inequality. According to Friedman’s permanent income hypothesis, in
fact, households prefer a smooth consumption flow, that is they choose to consume
a constant fraction of the permanent income because annual income is too volatile.
Assuming that income can be decomposed into a permanent component and a tran-
51.21∗∗∗
(5.596)
95
0.282
46.82∗∗∗
(6.382)
95
0.301
Mafia mean index
(1)
(2)
−40.13∗∗
−42.90∗∗
(17.24)
(16.94)
−0.213
−0.376∗
(0.173)
(0.207)
0.355
(0.232)
51.10∗∗∗
(5.614)
95
0.284
−18.89
(41.57)
(3)
−41.25∗∗
(17.38)
−0.205
(0.174)
58.19∗∗
(27.66)
95
0.283
−0.217
(0.885)
(4)
−41.73∗∗
(18.43)
−0.302
(0.386)
50.60∗∗∗
(5.489)
95
0.287
−17.71
(20.16)
(5)
−42.77∗∗
(17.68)
−0.193
(0.173)
51.32∗∗∗
(5.480)
95
0.303
−124.7
(84.73)
(6)
−44.09∗∗
(16.80)
−0.211
(0.170)
0.345
(0.876)
51.00∗∗∗
(5.562)
95
0.284
(7)
−37.81∗∗
(18.74)
−0.27
(0.238)
All results are based on the baseline sample of 5-year averages. Each cell reports the coefficient estimates and robust standard errors in the parenthesis. ∗∗∗,
∗∗, and ∗ denote significance of the regression coefficient at 1%, 5%, and 10%
Observations
R-squared
Constant
Part. in education and training
Total hours worked
Gross fixed capital formation
Long-run unemployment
Compensation of employees
Secondary Education level
Economic activity rate
Dependent variable
Regressors
Absolute Upward Mobility: Expected Rank
Table 16 The table presents cross-sectional regressions
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
233
42.59∗∗∗
(6.679)
95
0.27
37.63∗∗∗
(7.295)
95
0.288
Mafia mean index
(1)
(2)
−24.65∗∗
−26.69∗∗
(11.80)
(11.43)
−0.335∗∗
−0.500∗∗∗
(0.134)
(0.178)
0.347
(0.231)
42.30∗∗∗
(6.690)
95
0.271
−14.94
(41.43)
(3)
−25.22∗∗
(11.83)
−0.331∗∗
(0.134)
42.43
(26.98)
95
0.27
0.00515
(0.876)
(4)
−24.63∗∗
(11.93)
−0.332
(0.396)
41.67∗∗∗
(6.505)
95
0.273
−14.01
(19.68)
(5)
−25.87∗∗
(11.82)
−0.326∗∗
(0.132)
41.64∗∗∗
(6.426)
95
0.291
−123.4
(83.71)
(6)
−27.93∗∗
(11.60)
−0.339∗∗
(0.131)
0.517
(0.856)
43.06∗∗∗
(6.721)
95
0.273
(7)
−22.38∗
(12.17)
−0.410∗∗
(0.182)
All results are based on the baseline sample of 5-year averages. Each cell reports the coefficient estimates and robust standard errors in the parenthesis. ∗∗∗,
∗∗, and ∗ denote significance of the regression coefficient at 1%, 5%, and 10%
Observations
R-squared
Constant
Part. in education and training
Total hours worked
Gross fixed capital formation
Long-run unemployment
Compensation of employees
Secondary Education level
Economic activity rate
Dependent variable
Regressors
Absolute Upward Mobility: Q1Q5
Table 17 The table presents cross-sectional regressions
234
M. Battisti et al.
26.90∗∗∗
(9.656)
95
0.295
23.27∗∗
(9.017)
95
0.308
Mafia mean index
(1)
(2)
69.17∗∗
68.66∗∗
(27.02)
(27.51)
−0.319∗∗∗
−0.473∗∗
(0.112)
(0.185)
0.295
(0.239)
25.51∗∗
(9.999)
95
0.299
−28.05
(41.35)
(3)
72.71∗∗
(27.91)
−0.311∗∗∗
(0.113)
33.49
(27.04)
95
0.296
−0.231
(0.827)
(4)
71.42∗∗∗
(26.36)
−0.421
(0.389)
21.75∗∗
(9.851)
95
0.308
−30.5
(20.35)
(5)
81.25∗∗∗
(28.36)
−0.287∗∗∗
(0.109)
24.63∗∗
(9.667)
95
0.319
−131
(80.99)
(6)
75.93∗∗∗
(26.86)
−0.330∗∗∗
(0.112)
0.919
(0.814)
26.65∗∗∗
(9.341)
95
0.304
(7)
69.44∗∗
(26.65)
−0.426∗∗
(0.162)
All results are based on the baseline sample of 5-year averages. Each cell reports the coefficient estimates and robust standard errors in the parenthesis. ∗∗∗,
∗∗, and ∗ denote significance of the regression coefficient at 1%, 5%, and 10%
Observations
R-squared
Constant
Part. in education and training
Total hours worked
Gross fixed capital formation
Long-run unemployment
Compensation of employees
Secondary Education level
Economic activity rate
Dependent variable
Regressors
Relative Mobility
Table 18 The table presents cross-sectional regressions
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
235
28.57
(17.29)
95
0.172
22.33
(18.93)
95
0.184
Mafia rank index
(1)
(2)
164.2∗∗∗
163.3∗∗∗
(51.87)
(53.84)
−0.101
−0.365
(0.188)
(0.328)
0.506
(0.533)
29.48
(18.28)
95
0.172
18.45
(85.92)
(3)
161.8∗∗∗
(54.22)
−0.107
(0.193)
43.05
(62.19)
95
0.173
−0.507
(2.046)
(4)
169.1∗∗∗
(54.08)
−0.324
(0.925)
25.9
(18.16)
95
0.173
−15.81
(40.67)
(5)
170.4∗∗∗
(55.22)
−0.0845
(0.186)
24.97
(17.78)
95
0.19
−207.5
(145.3)
(6)
174.9∗∗∗
(53.24)
−0.118
(0.186)
−2.189
(1.776)
29.15
(17.66)
95
0.187
(7)
163.5∗∗∗
(52.59)
0.153
(0.304)
All results are based on the baseline sample of 5-year averages. Each cell reports the coefficient estimates and robust standard errors in the parenthesis. ∗∗∗,
∗∗, and ∗ denote significance of the regression coefficient at 1%, 5%, and 10%
Observations
R-squared
Constant
Part. in education and training
Total hours worked
Gross fixed capital formation
Long-run unemployment
Compensation of employees
Secondary Education level
Economic activity rate
Dependent variable
Regressors
Relative Mobility
Table 19 The table presents cross-sectional regressions
236
M. Battisti et al.
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
237
sitory component (see e.g., MaCurdy 1982) we would expect permanent shocks
(e.g., a technological shock that affects the need for unskilled workers) to affect
consumption and welfare because it harder to insure against them. In contrast, we
would not expect transitory shocks to affect consumption because individuals can
smooth those shocks by borrowing or using their assets. In practice, however, full
smoothing is infeasible due to the presence of borrowing constraints and other
imperfections of credit and insurance markets. For these reasons, we consider both
consumption and income inequality when we study their role in organized crime.
Interestingly, we find that consumption inequality much greater impact on organized crime than income inequality.
These results shed new light on the economic explanations of the rise and spread
of organized crime and suggest potential policy interventions. For example, the fact
that consumption inequality seems to matter more than income inequality, suggests
that what matters are differences in actual standards of living, which are better captured by consumption levels than by income levels as they depend on life-cycle
decisions (see e.g. Jappelli and Pistaferri 2010 for recent trends of income and consumption inequality in Italy). This directly points out a policy response in terms of
insuring adequate standard of living to the poor in order to reduce the existing disparities. The same holds for improving the socio-economic mobility, especially by
improving the perspective of those lagging behind the social ladder.
Conclusions
This study investigates the linkages between organized crime and inequality and
social mobility. In doing so we construct a novel dataset of organized crime at the
regional level based on two complementary approaches. The first one is based on
the methodology developed by Calderoni (2011) and the second one employs a
dynamic factor model. Additionally, we construct a wide range inequality as well as
social mobility measures. Our dataset is of general interest beyond the specific
application considered in the present study. Our main finding is that higher inequality can lead to higher organized crime development: in particular, the index that
better captures the effect of inequality is P90/P10. We also find that consumption
inequality performs better that income inequality as the relevant inequality measure.
Finally, we find that low socio-economic mobility displays a robust association with
organized crime development.
This is the first study analyzing the relationship between income inequality and
socio-economic mobility and organized crime. As such, it has obvious limitations.
In particular, although for the results on income inequality, especially those based
on GMM estimation, we can make a claim on causality running from inequality to
organized crime, for social mobility we cannot. The relationship between income
inequality and social mobility is a complex one, so that the dynamics of this relationship and its interaction with the emergence and spread of organized crime,
together with a precise identification of the significant channels at work remains to
be fully investigated. These topics are left to be explored in further research.
238
M. Battisti et al.
References
Acciari, P., Polo, A., & Violante, G. (2017). And yet, it moves’: Intergenerational mobility in Italy.
Mimeo NYU, New York.
Attanasio, O. P., & Pistaferri, L. (2016). Consumption Inequality. Journal of Economic Perspectives,
30(2), 3–28.
Bańbura, M., & Modugno, M. (2014). Maximum likelihood estimation of factor models on datasets with arbitrary pattern of missing data. Journal of Applied Econometrics, John Wiley & Sons,
Ltd., 29(1), 133–160. 01.
Barone, R., & Masciandaro, D. (2011). Organized crime, money laundering and legal economy:
theory and simulations. European Journal of Law and Economics, 32(1), 115–142.
Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data
models. Journal of Econometrics, 87, 11–143.
Caglayan, M., Flamini, A. and Jahanshahi, B. (2017). Organized crime and human capital DEM
working paper series nr. 143, Università di Pavia.
Calderoni F. (2011). Where is the Mafia in Italy? Measuring the presence of the Mafia across
Italian Provinces. Article published in Global Crime.
Chetty, R., Hendren, N., Kline, P., & Saez, E. (2014). Where is the land of opportunity? The geography of intergenerational mobility in the United States. The Quarterly Journal of Economics,
129(4), 1553–1623.
Coniglio, N. D., Celi, G., & Scagliusi, C. (2010). Organized crime, migration and human capital
formation: Evidence from the South of Italy. WP nr 28. Dipartimento: di Scienze Economiche
e Metodi Matematici-Universita di Bari.
Corak, M., & Heisz, A. (1999). The intergenerational earnings and income mobility of Canadian
men: Evidence from longitudinal income tax data. Journal of Human Resources, 34, 504–533.
Cowell, F. (2011). Measuring inequality. Oxford University Press. New York.
DIA. (2016). Relazione del Ministro dell’interno al Parlamento sull’attività svolta e sui risultati
conseguiti dalla Direzione Investigativa Antimafia. Roma. Available at: http://direzioneinvestigativaantimafia.interno.gov.it/page/relazioni_semestrali.html
Durlauf, S. (1996). A theory of persistent income inequality. Journal of Economic Growth, 1(1),
75–93.
Durlauf, S. (1999). The memberships theory of inequality: ideas and implications. In E. Brezis &
P. Temin (Eds.), Elites, Minorities, and Economic Growth. Amsterdam: North-Holland.
Durlauf, S. (2006). Groups, social influences, and inequality: A memberships theory perspective on poverty traps. In S. Bowles, S. Durlauf, & K. Hoff (Eds.), Poverty Traps. Princeton
University Press: Princeton.
Durlauf, S., & Seshadri, A. (2018). Understanding the great gatsby curve. NBER Macroeconomics
Annual, 32, 333–393.
Fajnzylber, P., Lederman, D., & Loayza, N. (2002). Inequality and violent crime. The Journal of
Law and Economics, 45(1), 1–39.
Fioroni, T., Lavezzi, AM and Trovato, G. (2017). Organized crime, corruption and poverty traps.,
in progress.
Franchetti, L. (1877). Condizioni politiche e amministrative della Sicilia (Vol. 1). G. Barbera.
Firenze.
Gambetta, D., & Reuter, P. (1995). Conspiracy among the many: The mafia in legitimate industries. In The economic dimensions of crime (pp. 99–120). UK: Palgrave Macmillan.
Garbinti, B., Goupille-Lebret, J., & Piketty, T. (2018). Income inequality in France, 1900–2014:
Evidence from Distributional National Accounts (DINA). Journal of Public Economics, 162,
63–77.
Güell, M., Pellizzari, M., Pica, G., & Rodríguez Mora, J. V. (2017). Correlating social mobility and
economic outcomes. Forthcoming Economic Journal.
Hertz, T. (2006). Understanding mobility in America. Center for American Progress Discussion
Paper.
Istat. (2010). Distribuzione del reddito e condizioni di vita in Italia.
Socio-Economic Inequalities and Organized Crime: An Empirical Analysis
239
Jappelli, T., & Pistaferri, L. (2010). Does consumption inequality track income inequality in Italy?
Review of Economic Dynamics, 13(1), 133–153.
Jong-Sung, Y., & Khagram, S. (2005). A comparative study of inequality and corruption. American
Sociological Review, 70(1), 136–157.
Lavezzi, A. M. (2014). Organised crime and the economy: a framework for policy prescriptions.
Global Crime, 15(1–2), 164–190.
Macours, K. (2010). Increasing inequality and civil conflict in Nepal. Oxford Economic Papers,
63(1), 1–26.
MaCurdy, T. E. (1982). The use of time series processes to model the error structure of earnings in
a longitudinal data analysis. Journal of Econometrics, 18(1), 83–114.
Moench, E., Ng, S., & Potter, S. (2013). Dynamic hierarchical models. Review of Economics and
Statistics, 95, 5.
Riccardi, M., Soriani, C., & Giampietri, V. (2016). Mafia infiltration in legitimate companies in
Italy. Organised Crime in European Businesses, 119.
Roodman, D. (2009). How to do xtabond2: An introduction to difference and system GMM in
Stata. Stata Journal, 9(1), 86–136.
Savolainen, J., Aaltonen, M., Merikukka, M., Paananen, R., & Gissler, M. (2014). Social mobility
and crime: Evidence from a total birth cohort. British Journal of Criminology, 55(1), 164–183.
Sharkey, P., & Torrats-Espinosa, G. (2017). The effect of violent crime on economic mobility.
Journal of Urban Economics, 102, 22–33.
Sunde, U., Cervellati, M., & Fortunato, P. (2008). Are all democracies equally good? The role of
interactions between political environment and inequality for rule of law. Economics Letters,
99(3), 552–556.
The Criminal Careers of Italian Mafia
Members
Ernesto U. Savona, Francesco Calderoni, Gian Maria Campedelli,
Tommaso Comunale, Marco Ferrarini, and Cecilia Meneghini
Introduction
This article draws its main results from the study of the criminal careers of Italian
mafia members, which is one of the innovative studies in project PROTON. The
analysis of criminal careers of mafia members covers the years from 1982 up to
2017. A long period that has allowed the researchers involved in the study to analyze the data related to the careers of mafiosi across two different levels of analysis:
their internal statistical significance and the transformations of mafias in Italy. This
article deals with the first level, introducing the second one in the conclusions and
postponing it to a larger analysis and publication.
After outlining why the analysis of the criminal trajectories of Italian Mafiosi
could contribute to the development of the theoretical framework of criminal
careers, this study explains the secondary data sources used, how these data could
fit with Blumstein et al.’ (1986) theoretical framework and how five of the six
Blumstein parameters have been operationalized through the available data on convicted mafia offenders in Italy. The chapter follows a Macro/Meso/Micro approach,
using the parameters to look at the criminal career of “Mafiosi” from different
points of view: the macro level aims at identifying the overall patterns of evolution
comparing them with existing knowledge on other serious offenders (section “The
Criminal Trajectories of mafiosi”). The meso level compares the types of mafias
“Mafiosi” belong to, exploring if and what are similarities and differences (section
“Similarities and Differences of Italian mafiosi Across Their Criminal
Organizations”). The micro level focuses on individuals, to examine the temporal
E. U. Savona (*) · F. Calderoni · G. M. Campedelli · T. Comunale · C. Meneghini
Università Cattolica del Sacro Cuore and Transcrime, Milan, Italy
e-mail: ernesto.savona@unicatt.it
M. Ferrarini
Business Intelligence, Milan, Italy
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_10
241
242
E. U. Savona et al.
and spatial dimension of mafia involvement, assessing what factors influence the
recruitment of mafia offenders into the organization (section “When, Why and
Where Do mafiosi Enter in Their Criminal Organization?”). Discussion and conclusions follow (section “Discussion and Conclusions”), framing the results of this
study in a better understanding of the criminal careers of serious criminals, as Italian
“Mafiosi” are, and the mechanisms that have driven their behaviors.
hy Italian Mafiosi? What We Can Learn from Their
W
Criminal Trajectory?
The demand for the dataset of the criminal careers of Italian mafiosi, as outlined in
the next section, was originated by the assumption that analyzing all the data of
Italian mafiosi convicted between 1982 and 2017 could have helped to explain the
processes of recruitment of mafia members in their organizations. The necessary
process of anonymization and the relevant missing values of many significant variables have moved these expectations towards two different research questions that
are collateral to the issue of recruitment. The former related to how the criminal
trajectories of Italian mafiosi fit with the theoretical framework of criminal career
conceived by Blumstein in 1986. The second, using Blumstein’s patterns, to how
these trajectories can explain the transformations of organized crime in Italy across
more than 30 years. In both cases the analysis of criminal trajectories of Italian
mafiosi can reveal something to researchers and policy makers for now and the
future. It could help, with better data, to refine the different trajectories of mafiosi
across the four criminal organizations in Italy, understanding similarities and differences, and discuss and interpret the transformations of mafias in Italy.
Merging Two Datasets: Variables and Their Robustness
This study relies on a new original dataset, the Proton Mafia Members dataset
(PMM). The PMM originates from two datasets provided by the Italian Ministry of
Justice: the Criminal Records Registry (Casellario dataset) and the Prison
Administration Department dataset (henceforth DAP dataset). Formal agreements
with the Ministry of Justice made the data available and guarantee the anonymity of
all individuals in compliance with current data protection and privacy regulations.
The Casellario dataset provided information on the criminal records for individuals
convicted for mafia offenses between 1982 and March 2017.1 The dataset included
The research team considered as mafia offenses the crime of mafia association and other offenses
aggravated by the mafia method (i.e.: Articles 416bis, 416ter, 418 of the Italian Criminal Code, and
special laws 575/1965, Art. 7 special law 152/1991 and Art. 7 special law 203/1991).
1
The Criminal Careers of Italian Mafia Members
243
182,867 offenses, with detailed information on the legislative source (e.g. Italian
Criminal Code), article number, and paragraph of the violated provisions. The
research team classified the offenses into 31 categories according to the characteristics of the crimes. The most recurrent categories were weapons and explosives (felonies), extortion, theft, drug-related crimes (i.e. illicit drugs production, trafficking
and selling), fencing, and mafia association (Savona et al. 2017a). The DAP dataset
contained socio-demographic variables, arrest and imprisonment data, information
related to the type mafia association (e.g. Cosa Nostra or ‘Ndrangheta), and the role
that mafia members held within their group (e.g. affiliate, boss).
The PMM dataset consists of a sample (n = 11,144) of mafia members. It provides socio-demographic data, information on the type of mafia group (i.e. mafia
association) and the role, as well as information on mafia members’ criminal career
(Table 1). In the PMM dataset each observation is a unique alphanumeric ID. The
research required several data cleaning and data transformation procedures to detect
and correct inaccurate records. Data preparation included imputation for missing
data to set up relevant variables for the analyses (Savona et al. 2017a). For instance,
the treatment of mafia association led to a more robust variable with missing values
reduced from 43% to 5%. Conversely, few other variables remained with an important amount of missing values, as the role of mafia members within their mafia
group (Table 1). The research team carefully considered this aspect to assess the
validity and inference of the analyses performed (2017).
The analysis of the criminal careers of Italian mafias’ members relies on the
conceptual framework of the criminal career paradigm developed by Blumstein
et al. (1986). A criminal career is defined as “the longitudinal sequence of offenses
committed by an offender who has a detectable rate of offending during some
period” (Blumstein et al. 1988, p. 2). Longitudinal data on offending throughout the
life course represent the ideal data to study the criminal career of individuals
(Farrington 1992). Longitudinal patterning of criminal activity can be investigated
using self-report and/or official data. Despite some limitations (see Piquero et al.
2003), official records can provide precise information about the timing of offences
through conviction data. This study seeks to investigate the criminal career of mafia
members through the analysis of unique conviction data contained in the PMM
dataset.
The analysis operationalizes five of the six parameters related to the criminal
careers framework, namely: participation, duration, frequency, specialization, and
escalation (Savona et al. 2017a).2 In the context of offending, participation refers to
“the proportion of a population who are active offenders at any given time”
(Blumstein et al. 1988, p. 3). Active offenders include both new offenders and existing offenders, i.e. individuals with previous criminal convictions (see Piquero et al.
2003). The analysis of the PMM dataset led computing two different participation
measures. External participation is the ratio between the number of mafia members
The intermittency parameter was discarded, as available data prevent identification of the prison
times served by mafia members.
2
244
E. U. Savona et al.
Table 1 Descriptive statistics for some variables in the PMM dataset (n = 11,144)
Variable
Sex
Country of birth
Religion
Education
Mafia association
Role
Mean age at first crime
Mean age at first mafia crime
Mean age at first arrest
Mean number of committed
crimes
Mean number of crime
categories
Mean number of violent
crimesa
Criminal careers’ parameter
Participation (internal)
Statistics
98% are males
98% are born in Italy
99% are Catholic
82% have between 5 and 8 years of education
37% belong to the Sicilian Mafia
65% are associates, 27% underbosses, 7% bosses
24.61
34.30
37.85
16.40
Missing
0%
0%
12%
18%
5%
56%
3%
16%
0%
0%
7.34
0%
3.42
0%
14.29 (avg. % of active offenders committing crimes
per year)
Duration
14.95 years
Frequency
1.52 crimes per year
Specialization (diversity index) 0.75
Escalation by age
4.84
Escalation by crime number
15.33
Most frequent categories (n = 182,867)
Weapons and explosives
17%
(felonies)
Extortion
9%
Mafia association
8%
Theft
7%
Murder
6%
NA
2%
11%
4%
25%
25%
NA
NA
NA
NA
NA
Violent crimes include murder, extortion, assault and violent offenses, and robbery
a
committing at least one crime in 1 year and the resident population aged 14 and over
living in the four Italian regions with a traditional mafia presence (i.e. Sicily,
Campania, Calabria, and Apulia). Internal participation is the ration between the
number of mafia members committing at least one crime in any year and the number
of potentially active members in the PMM dataset, i.e. individuals aged between 14
and 90 in the same year. The duration of a criminal career is the “length of time that
an offender continues to commit crimes once beginning an active criminal career”
(Rhodes 1989, p. 3). In the PMM dataset, the duration is the time span between the
first and the last recorded offence of each individual. The individual offending frequency refers to “the average number of crimes committed per year by active
offenders” (Farrington et al. 2016, p. 339). Among criminal offenders, ­specialization
The Criminal Careers of Italian Mafia Members
245
is “the tendency to commit the same type of crime” (Kyvsgaard 2002, p. 147). The
study assessed the specialization by resorting to the Diversity index.3 Finally, the
escalation parameter is the tendency to move to more serious offense types
(Blumstein et al. 1986). The study computed seriousness by considering the average
statutory penalty for each crime committed, due to missing data on the actual
imposed sentences. The escalations parameter is the coefficient of a linear regression of the seriousness of the crimes on a temporal measure. Two different temporal
measures were used: the escalation by age regresses the seriousness on the offender’s age, thus measuring the escalation as individuals age. The escalation by crime
regresses the crime seriousness on the progressive number offence in an individual’s
criminal career, focusing on the escalation as subsequent crimes are committed.
As in other studies relying on administrative data and criminal records, the operationalization of the five parameters required a long process of data cleaning.
Furthermore, the measurement of the variables (participation, duration, frequency,
specialization, and escalation) through available data implied complex choices that
have been explained in this section. The data have been used at the best of their possibilities and, the analytical framework that follows in the next section has been
chosen as the best framework for combining the phenomena analyzed, the research
questions and the data, including their limitations.
he Analytical Framework: Macro, Meso, Micro Dimensions
T
of the Criminal Careers of Italian Mafiosi
For the purpose of this study, we consider members of the Italian mafias all individuals who were convicted for a mafia offence (see footnote 1 above) and whose
conviction if final according to the Italian criminal justice system. This criterion is
certainly very restrictive, as it may exclude e.g. individuals with pending criminal
proceedings or whose cases were dismissed due to statute of limitations.
Nevertheless, the selection provides a clear threshold for delimiting the sample
population. The analysis of the criminal careers of Italian mafia members relies on
a three-level approach that focuses on their macro, meso, and micro dimensions.
The macro dimension analyses the whole sample of mafia members, focusing on
the identification of specific trajectories through Group Based Trajectory Modelling.
The goal of the first level is to identify overall patterns of evolution in the career of
3
m =1
The diversity index (DIi) for individual i is defined as DI i = 1 − ∑ pm( ) ∗ pm( ) , where m = 1, 2…31
31
i
i
are the 31 crime categories identified in the dataset and pm( ) is the proportion of offenses committed by individual i in the crime category m. The diversity index can be interpreted as the probability
that any two offenses drawn randomly from an individual’s set of offenses belong to two different
crime categories (Piquero et al. 1999). When DIi=0, offender i is completely specialised on one
type of crime. Conversely, a value of the index approaching 1 indicates that the offender engages
in a diversity of crime categories. Given the number of crime categories identified in the dataset,
the maximum value of the diversity index in this analysis is (31-1)/31 = 0.97.
i
246
E. U. Savona et al.
Italian mafiosi and compare them with existing knowledge on other serious offenders. The meso dimension focuses on the different types of Italian mafias, namely,
Camorra, Sicilian Mafia, ‘Ndrangheta, and Apulian mafia. Notoriously, Italy hosts
several mafias which differ under many aspects. The most popular Italian mafias are
the Sicilian Cosa Nostra, the ‘Ndrangheta (originating from the Calabria region),
the Camorra (from the Campania region) and the Apulian Sacra Corona Unita. Each
mafias has peculiar historical, cultural and organizational features (Berlusconi
2014; Paoli 2014). While Cosa Nostra and the ‘Ndrangheta are a confederation of
different groups, the Camorra does not constitute a confederation: “Camorra” is the
name employed to refer to independent gangs and criminal groups located in the
province of Naples and in the surrounding areas (Paoli 2014). The Sacra Corona
Unita was active in Apulia since the early 1980s, but the arrest of most of its members at the beginning of the current century led to its decline and to the emergence
of a variety of other criminal groups currently active in the region (Direzione
Investigativa Antimafia 2002; Europol 2013). Given the differences across the
mafias, the meso dimension explores the possible discrepancies across the criminal
careers of members of different mafias. The micro level dimension analyses the
dynamics and mechanisms of mafia involvement and careers at individual level,
providing information on how mafia members’ careers evolve. It can shed light on
the specific characteristics of the criminal careers of mafia members driving their
recruitment into the mafias.
The analyses of each dimension employed different quantitative techniques,
namely group-based trajectory modelling, regression models, and statistical correlation, using the criminal careers parameters as fundamental variables in all the
analyses. The analysis exploited available social and demographic variables, in
addition to the criminal careers parameters calculated for each individual in the
whole population. Further details on the data sources, the data cleaning and the
computation of the parameters of the criminal careers are available in the research
report of Project PROTON (Savona et al. 2017a).
While it is the first time that this type of analysis if conducted on the Italian
Mafiosi, relying on a large and original dataset provided by the Italian Ministry of
Justice, the data have a few limitations. First, the analysis focused only on final
convictions, to precautionarily include only crimes whose commission was finally
established by the courts. In the Italian criminal justice system, a conviction is final
only when the deadline for filing an appeal in a higher court has expired, or when
there are no higher courts to appeal to. The law gives the defendant and the prosecution the possibility to appeal the judgement of the tribunals, and appeal decisions
can be further challenged before the Corte di Cassazione (the Italian supreme court
for civil and criminal cases). This process may take several years, depending on the
type of offence and the resources available to the parties. Given the seriousness of
the possible penalties, mafia-related criminal proceedings are often likely to go
through three separate decisions, and even more (e.g. the supreme court may quash
an appeal judgement and order a new appeal trial, which might be further challenged before the supreme court). As a result, the most recent period in the sample
is underrepresented as many proceedings are likely still pending. Therefore, our
The Criminal Careers of Italian Mafia Members
247
results are mostly representative of the mafia members’ population convicted
between the 1980s and the 2000s and have very limited application to the most
recent years. Second, the PMM lacks some important information which may be
useful for examining the criminal careers of mafia members. For example, there is
no information about the imposed penalties for each crime and consequently it is
impossible to reliably determine the prison time served by each individual. Third,
the PMM does not provide data on co-offending (whether two or more criminals
were convicted for the same offence), nor on the affiliation to specific mafia family
or clan. Due to this lack of information it was impossible to explore the criminal
career patterns in relation to the dynamics of specific criminal groups.
The Criminal Trajectories of Mafiosi
The interest over the developmental evolution of criminal involvement has attracted
the attention of many scholars in the last decades. Researchers have applied Group
Based Trajectory Modelling (henceforth GBTM) as the most common statistical
technique for investigating individual and collective trajectories of crime across
criminal samples of different nature. In a landmark paper, Laub and co-authors
(1998) have relied on longitudinal data collected by Glueck and Glueck (1968) to
analyze the criminal trajectories of juvenile delinquents. In their study, they have
demonstrated how desistance from crime is facilitated when individuals engage in
quality marital bonds, moreover highlighting how this positive influence is gradual
and cumulative over time. Similarly, Blokland and Nieuwbeerta (2005) have proved
how marriage and other life circumstances can have individual effects on the development of criminal trajectories over time, using a sample of Lussier (2010) and
colleagues have instead focused on sexual crimes. They have studied a quasi-­
population of convicted adults in Quebec and detected four different offending trajectories. Their analysis stressed the difference across groups in terms of age of
onset, frequency, diversity and specialization, thus contesting the idea that sex
offenders have stable and linear profiles. Employing a nationally representative
sample of Dutch men convicted in 1977, van de Rakt et al. (2012) have concentrated
on the effects of paternal imprisonment on criminal trajectories of children, finding
a mixed relationship between fathers’ imprisonment and child convictions. In fact,
paternal imprisonment does not alter the shape of development of a criminal career
but it alters the height of a criminal trajectory. Additionally, authors found that children whose fathers were convicted before they were 12 years old had a higher probability of imprisonments in the age between 18 and 30.
Applications of GBTM to more serious forms of crime, as white-collar crime or
organized crime, is rarer. Among the few examples, van Onna and colleagues analyzed the criminal career profiles of a sample of white-collar criminals (2014),
showing four different types of criminals with consistent patterns also in terms of
selection offence characteristics and sociodemographic background. Van Koppen
et al. (2010) investigated the existence of specific longitudinal trajectories for a
248
E. U. Savona et al.
sample of 854 Dutch organized criminals. Also in this case, the study identified four
distinct groups holding specific features. Both these studies shed light on the existence of late onset mechanisms for serious offenders. Specifically, Van Koppen and
co-authors point out how the insufficient attention on late onset can be explained by
the high attention given to youth samples in criminal trajectories literature and by
the interest on high volume crime in which adult-onset is scarce. Contrarily, studies
on white collar and organized crime highlight the presence of subgroups of individuals who start to engage in crime in their adult age following dynamics which
Kleemans and de Poot tried to explain through the theoretical framework of social
opportunity structure (2008). However, in spite of the wide interest of criminologists for the study of criminal careers over time, no research existed on the trajectories of mafia members to date. To the best of our knowledge, this is the first
comprehensive analysis of the criminal trajectories of Italian mafia members.
In this study, we apply GBTM to the criminal career of Italian Mafiosi. Group
Based Trajectory Modelling (henceforth GBTM) is a statistical technique that aims
to detect different subpopulations that share common offending trajectories over
their careers and detect significant patterns (Jones and Nagin 2013). Technically,
GBTM is a specialized form of finite mixture modelling employed to analyze developmental trajectories measuring a specific outcome over certain time periods, and
has been extensively applied in Criminology (see Nagin, 2016). In the present study,
the outcome (i.e. the dependent variable) is represented by the number of offenses
committed by an individual at a given age. Relying on the age as temporal dimension (instead of the years) allows to position all the individuals in the dataset at the
same starting point. We adapted the PMM data for the use of GBTM through two
strategies. First, the time-span covers crimes committed by Italian mafiosi from age
14 to age 60.4 Second, we censored the maximum number of offence per year setting it at 15. This figure is reasonably high, as it is close to mafia members’ average
number of offences across the whole career, and allows for a correct computation of
the models.5
The analysis of the trajectories within the mafia members population (n = 11,144)
led to the identification of five different groups that reflect distinct developmental
and criminal patterns (Fig. 1). We labelled the trajectories as High Frequency
Offenders (HFO), Low Frequency Offenders (LFO), Moderate Persistence Offenders
(MPO), High Persistence Offenders (HPO), and Early Starters (ES).
Each trajectory group holds specific features (Table 2). High Frequency Offenders
(10.33% of the sample) start early and reach their criminal peak at age 30. As
Some mafia members in the PMM still commit offences after age 60. However, considering the
lack of information on the death date and the technical problem of high singularity of the matrix
(the high occurrence of zero offences for the years from age 55 to age 75), the dataset was censored
to age 60.
5
Indeed, at a first stage the absolute variance of the entire matrix was too high for the correct computation of the models. Some individuals committed more than 100 offences in a single age/year,
while zeros accounted for 75% of the cells of the matrix after the computation (98% before temporal censoring).
4
The Criminal Careers of Italian Mafia Members
249
Fig. 1 Criminal trajectories of mafia members (n = 11,144)
Table 2 Descriptive statistics of mafia members’ trajectories
Group
High frequency
offenders
Low frequency
offenders
Moderate
persistence
offenders
High persistence
offenders
Early starters
Overall
N of
individuals
1151
(10.33%)
3803
(34.13%)
2408
(21.61%)
961
(8.62%)
2821
(25.31%)
11,144
(100.00%)
Avg age at Avg N of Avg
first crime crimes
duration
20.52
39.51
17.68
Avg
frequency
3.21
Avg
seriousness
80.63
32.66
5.59
10.79
0.94
125.63
21.00
19.71
20.00
1.13
71.75
22.78
33.27
29.06
1.34
65.66
19.72
13.01
10.01
1.83
86.38
24.61
16.41
14.95
1.53
94.23
s­ uggested by the label, they exhibit the higher frequency overall and they are also
the most active in general, showing the highest average number of crimes committed. Conversely, Low Frequency Offenders report the lowest frequency and average
number of crimes committed. Their criminal activity is low and almost stable for the
entire course of their lives. Nevertheless, they tend to commit few but very serious
offences, considering that their average seriousness is the highest overall. Moderate
Persistence Offenders reach their highest criminal activity at age 28, moderately
persisting over their careers. Indeed, they converge towards zero at age 50. Their
250
E. U. Savona et al.
duration is above the average and they also tend to engage in crime early. High
Persistence Offenders show a sinusoidal trend over time. The most notable feature
is that they tend not to desist in the adult age: they exhibit their criminal peak at age
45, when most mafia members are not active anymore (due to desistance or incapacitation). Consequently, High Persistence Offenders have on average the longest
careers, but tend to commit the least serious crimes overall. Finally, the Early
Starters are mafia members who start their criminal careers at the youngest age. At
age 23 they reach their criminal peak, rapidly followed by a steep decrease. Their
trajectory converges towards the zero early (age 38), probably as a consequence of
an incapacitation effect linked to the high risk of conviction that unexperienced
young members face in their early careers.
The results are mostly in line with previous literature (Onna et al. 2014; Van
Koppen et al. 2010). Indeed, both studies identified the same number of groups (i.e.:
4) with similar trajectories shapes, with at least one group of low frequency offenders. Moreover, an early onset group is often related to higher frequencies while low
frequency offenders are associated to a later onset age. These findings point out
how, regardless of the specific type of criminal activity, mafia member seem to follow developmental trajectories similar to other serious criminals. The explanation
for this result may be found into future research on the underlying mechanisms that
push individuals to criminal activities at certain stages of their lives.
imilarities and Differences of Italian Mafiosi Across Their
S
Criminal Organizations
The research team assessed the differences across the careers of the members of the
four main types of Italian mafias: the Sicilian Mafia, the Camorra, the ‘Ndrangheta,
and the Apulian mafia. These four mafias differ in terms of criminal interests, organizational structure, modus operandi and reactions against the State action. Such
differences might cause the criminal career of their mafia members to follow distinctive paths, thus justifying the analysis of mafia members’ criminal careers at the
meso level (by type of mafia organization).
To explore whether the criminal careers of members belonging to different
mafias are significantly different, the analysis relied on the mafia association
variable. Originally, this variable was missing for 43.07% of mafia members.
However, the analysis exploited information on the mafia member’s province of
birth and on the region where mafia organizations are rooted to significantly
reduce the amount of missing data. First, mafia organizations were grouped by
Italian region.6 Second, individuals with a missing mafia association value but
A new variable (mafia association aggregated) was created by grouping the different values of
mafia association by Italian region. This resulted in five categories for mafia association aggregated: “‘Ndrangheta” (grouping members of ‘Ndrangheta and Other Mafias Calabria), “Sicilian
Mafia” (grouping members of Cosa Nostra, Stidda and Other Mafias Sicily), “Camorra” (consist6
The Criminal Careers of Italian Mafia Members
251
who were born in provinces with an established and predominant mafia group
were imputed this mafia group as mafia association.7 For example, 99.49% of
mafia offenders born in the province of Naples (and with a non-missing mafia
association value) are Camorra members, while 99.67% of offenders born in the
province of Reggio Calabria are ‘Ndrangheta members. 1314 mafia offenders
born in Naples and with a missing mafia association value were imputed
“Camorra” as mafia association, while 700 offenders born in Reggio Calabria
and with missing mafia association were coded as ‘Ndrangheta members.
Overall, the imputation reduced missing values from 43.07% (n = 4800) to 5.35%
(n = 596), thus allowing a more robust analysis of the similarities and differences
of the criminal careers of members of different mafia organizations.
The analysis of the trend in the number of offenses committed by members of
different mafia organizations reveals the presence of some meso-level differences
(Figs. 2 and 3). The analysis focuses on the number of crimes committed after 1982,
the year in which the mafia association crime was introduced in the Italian Penal
Code.8 In each year considered, most recorded offenses are committed by either
Camorra or Sicilian Mafia members. Despite the evidence on the dimension and
expansion of the ‘Ndrangheta in recent years (Ciconte 2014), crimes committed by
this mafia organization represent a minor share throughout the years 1982–2016.
This could be a consequence of the family structure characterizing the ‘Ndrangheta,
which leads to very few pentiti and guarantees secrecy and power to this mafia organization (Serenata 2014). At the same time, the underestimation by the Italian State
of the power and influence of the ‘Ndrangheta allowed its expansion abroad, especially when law enforcement authorities had their eyes on the Sicilian Mafia after
the murder of two anti-mafia judges in 1992 (Ciconte 2014).
Between 1984 and 1997, the Sicilian Mafia was the mafia organization registering more final convictions, while from 1997 onwards the share of offenses committed by members of the other mafias (especially Camorra members) acquires a
greater importance. This finding corroborates the view that the intensification of law
enforcement action that followed the “terrorist” mafia attacks in 1992–1993 led to a
decline in the power of the Sicilian Cosa Nostra (Paoli 2008). Indeed, the trend in
ing of Camorra members), “Apulian Mafia” (grouping members of Sacra Corona Unita and Other
Mafias Apulia) and “Mafia Lucana” (consisting of members of Mafia Lucana). Results for “Mafia
Lucana” are not presented in the following analysis as only 28 offenders in the dataset are members of this mafia organization, thus impeding to draw comparable conclusions on their criminal
career’s patterns.
7
For each Italian province where the number of mafia members born in that province was higher
than 100, the relative frequencies of the mafia groups (values of mafia association aggregated)
were calculated. If the relative frequency of a mafia group within a province was greater or equal
than 85%, that value was imputed into a new variable named mafia association aggregated clean
for all individuals born in that Italian province.
8
The sample includes n = 129,571 offenses out of the total 182,867 offenses in the PMM dataset.
This discrepancy is due to two reasons: firstly, offenses committed before 1982 or with missing
year are excluded from these charts. Secondly, offenses with a missing value for the mafia association variable (after the data imputation) are excluded as well.
252
E. U. Savona et al.
Fig. 2 Percentage of offenses committed by mafia association (on the total number of offenses
committed by mafia members, n = 129,571)
Fig. 3 Number of offenses committed by mafia association per year (n = 129,571)
the total number of offenses committed per year by the Sicilian Mafia (Fig. 3) displays a marked peak in 1991 followed by a sharp decline in the following years.9 On
The drop in the number of crimes committed by all mafia organizations in more recent years
(approximately from 2007 onwards) might be influenced by the fact that the dataset contains information on mafia members’ final convictions. In Italy, proceedings for mafia-related offenses often
last many years before reaching a final judgment. For example, the analysis of the data from the
9
The Criminal Careers of Italian Mafia Members
253
Table 3 Descriptive statistics of the criminal career’s parameters and variables in the four Italian
mafia organizations
Participation (internal)∗
Participation (external)∗
Duration
Frequency
N violent crimes
Specialization
Mean crime seriousness
Escalation (by crime number)
Escalation (by age)
Number of individuals
Type of mafia organization
Apulian mafia ‘Ndrangheta
18.81%
11.17%
0.0083%
0.0104%
15.46
13.55
(9.81)
(11.46)
1.71
1.42
(1.53)
(1.79)
2.91
2.16
(4.77)
(4.01)
0.81
0.72
(0.10)
(0.14)
74.45
107.28
(37.59)
(60.09)
13.90
10.95
(36.44)
(47.64)
5.80
3.39
(18.36)
(20.80)
1625
1710
Camorra
15.17%
0.0096%
15.40
(10.55)
1.53
(2.07)
3.83
(6.16)
0.77
(0.12)
86.73
(44.89)
16.85
(39.49)
5.52
(17.51)
3234
Sicilian Mafia
13.11%
0.0120%
15.00
(11.09)
1.46
(2.02)
3.78
(7.34)
0.74
(0.14)
104.76
(54.92)
16.84
(44.43)
4.34
(17.35)
3951
Notes: standard deviations reported in parenthesis. aFor internal and external participation, the
average participation over time (1982–2016) is shown. For all the other variables, the average
across individual level values is presented. ANOVA tests have been run for all the individual level
variables, revealing statistically significant differences (at the 0.1 per cent level) across different
mafia organizations for all of them
the other hand, the number of crimes committed by Camorra and Apulian mafia
members exhibits a similar peak in the early 1990s but remains quite sustained in
the following years.
The variables and parameters describing mafia members’ criminal careers present statistically significant differences in their mean values across mafia organizations (Table 3). Nonetheless, the distributions of the parameters in the four mafia
organizations are visually very similar (Savona et al. 2017a). Italian mafia organizations share some specific features that distinguish them from other types of criminal
organizations, such as the ability to control legitimate markets and the exertion of
political dominion over the areas of settlement (Paoli 2014). However, the documented differences in the criminal careers of members of different Italian mafia
organizations seem to underline some disparities in the criminal interests and modus
operandi, which likely relate to the historical development of such criminal
organizations.
Casellario shows that on average Italian courts issue definitive judgments (i.e. irrevocable sentences) 6 years after the commission of the mafia association crime.
254
E. U. Savona et al.
A general analysis of the variables and parameters defining mafia members’ criminal
careers in the four main Italian mafia organizations allows to grasp some of the underlying differences across these organizations. On average, Apulian mafia members exhibit
the most intense criminal careers: they have the longest careers, the highest offending
frequency and they are the most versatile offenders. They also display the highest mean
internal participation, indicating that on average a higher share of Apulian mafia members are active offenders over the period 1982–2016. Moreover, Apulian mafia members
exhibit the highest escalation by age, while Camorra and Sicilian Mafia members appear
to have on average the highest escalation by crime number. Despite presenting such
intense criminal careers, Apulian mafia members commit on average less serious
offences compared to members of all the other mafia organizations. The Apulian mafia
experienced the greatest expansion between the 1980s and 1990s, after which it faced
many arrests, trials and defections, which caused it to implement a strategy of violent
conflict against the State and other criminal organizations (Massari 2014). Moreover,
contrary to other mafia organizations, recruitment procedures in the Apulian mafia are
not based on biological-parental principles, causing offenders who seek to gain a leading
role in the organization to resort to violence, murder and other coercive tools to move up
the command hierarchy and maintain their position of power (Massari 2014).
The criminal careers of members of other mafia organizations seem to evolve differently. ‘Ndrangheta and Sicilian Mafia members are the most serious offenders in
terms of the average seriousness of committed crimes. Compared to members of
other mafia organizations, ‘Ndrangheta members have shorter careers and commit
crimes at lower frequency, but the average seriousness of committed offences is the
highest (despite not focusing on the perpetration of violent crimes). Their lower escalation values suggest that ‘Ndrangheta members commit more serious offences since
the early years of their criminal career. On the other hand, Camorra and Sicilian
Mafia members have on average more violent careers. The mean criminal career’s
duration and frequency for members of the Camorra and Sicilian Mafia is close to the
overall average, but the mean number of violent crimes committed is especially high.
It is hard to infer conclusions from this analysis where similarities and difference
across the four criminal organizations say very little. Further exploration with more
specific examination of the groups and better data may provide better insights into
specific characteristics of criminal careers within specific types of mafias. Individual
patterns as outlined in the next section could contribute to this analysis.
hen, Why and Where Do Mafiosi Enter in Their Criminal
W
Organization?
From a micro-level perspective, the study investigates the temporal and spatial
dimension of mafia involvement, trying to determine what factors influence when
and where mafia offenders enter in their criminal organization. Whereas more information is available as regards the age of entrance into the mafia organization, less
The Criminal Careers of Italian Mafia Members
255
can be inferred with respect to the spatial dimension of mafia involvement. A close
examination of the differences between “early recruits” and “late recruits” allows to
analyze some of the factors leading to mafia involvement, thus providing an exploratory answer as to why offenders enter in their mafia organization.
In line with previous studies on organized crime offenders (Francis et al. 2013;
Kleemans and de Poot 2008), the criminal career of Italian mafiosi tends to start at
a quite late age compared to what has been found in samples of common criminals:
on average, Italian mafia members commit their first crime at the age of 25 (median
age is 22). However, their entrance in the mafia organization (marked by their first
mafia association offense) usually occurs at an even later stage, being on average at
the age of 34 (median age is 33).10
At the individual level, the age of entrance into the mafia organization varies
significantly across the sample. While some members join the mafias as young
adults, other enter at later ages. To explore the characteristics influencing an early or
a late entrance into the mafia, the sample has been divided into three groups according to the distribution of the variable for the age of entrance into the mafia. More
specifically, offenders whose age at mafia entrance is lower or equal than the 25th
percentile of the distribution (27 years old) are classified as “early recruits”, while
offenders who entered the mafia at an age equal or higher than the 75th percentile
of the distribution (40 years old) are identified as “late recruits”. Drivers of early and
late recruitment are assessed through multinomial logistic regressions where the
dependent variable is the categorical variable dividing the sample in “early recruits”,
“late recruits” and offenders who joined the mafia at an “average” age (the baseline
category) (Table 4).11 In the first specification (the first and third column of Table 4),
the included predictors are the criminal career’s parameters, the number of violent
crimes committed and the mean seriousness of committed crimes. In the second
specification (the second and fourth column of Table 4), the model aims to detect
whether early and late recruits committed different types of crimes before entering
the mafia organizations. For this purpose, the included predictors are the number of
crimes committed under certain groupings of the crime categories.12 Moreover, all
specifications include the years of education as regressor and control for the year of
birth and the type of mafia association.
The year of the first mafia association offence is employed as a proxy for the timing of entrance
into the mafia organization. This methodological choice may present some flaws linked to the
imprecision of the year of crime variable and to the possibility that mafia offenders committed
some mafia association crimes at younger ages that might have gone undetected by law enforcement authorities. Nonetheless, some robustness checks relying on building time-buffers that
“move” the year of recruitment to 1, 2 and 3 years before and after the year of the first mafia association crime showed that the choice of this last variable as year of entrance into the mafia organization is fairly robust (Savona et al. 2017a).
11
Most results are robust to the change in the age limit defining “early” and “late” recruits (Savona
et al. 2017a).
12
Crimes have been grouped through a Principal Component Analysis that enabled to identify
relevant correlations across the 31 crime categories and reduce them into six groupings (Savona
et al. 2017a).
10
256
E. U. Savona et al.
Table 4 Drivers of early and late recruitment: multinomial logistic regressions
‘Ndrangheta
Camorra
Sicilian Mafia
Mean serious. pre MO
Duration pre MO
Frequency pre MO
Escalation pre MO
Specialization pre MO
N violent crimes pre MO
N long conviction crimes pre MO
N robbery, theft and related crimes
pre MO
N arson and assault crimes pre MO
N drug-related crimes pre MO
N admin, financial, road traffic
crimes pre MO
N white-collar crimes pre MO
N associative crimes pre MO
Years of education
Year of birth
Intercept
Pseudo R2
Base outcome is the average recruitment age (28–
39 years old)
Early recruits (<28 years
Late recruits (>39 years
old)
old)
−0.275
−0.486
0.474
0.174
(0.251)
(0.156)∗∗
(0.226)∗
(0.168)
0.270
−0.047
0.157
−0.089
(0.172)
(0.122)
(0.191)
(0.151)
−0.146
−0.455
0.645
0.434
(0.172)
(0.113)∗∗∗
(0.181)∗∗∗ (0.140)∗∗
−0.007
0.004
(0.002)∗∗
(0.002)
−0.492
−0.358
0.275
0.168
(0.028)∗∗∗ (0.014)∗∗∗
(0.015)∗∗∗ (0.008)∗∗∗
0.115
0.191
(0.070)
(0.110)
0.003
−0.011
(0.003)
(0.005)
1.961
−1.588
(0.504)∗∗∗
(0.442)∗∗∗
0.014
−0.069
(0.025)
(0.020)∗∗∗
0.001
−0.021
(0.009)
(0.008)∗∗
0.158
−0.197
(0.016)∗∗∗
(0.018)∗∗∗
0.132
−0.076
(0.030)∗∗∗
(0.033)∗
−0.119
0.090
(0.039)∗∗
(0.031)∗∗
0.071
−0.035
(0.029)∗
(0.013)∗∗
−0.311
0.170
(0.077)∗∗∗
(0.054)∗∗
0.423
−0.292
(0.123)∗∗∗
(0.105)∗∗
−0.075
−0.080
0.105
0.090
(0.027)∗∗
(0.017)∗∗∗
(0.022)∗∗∗ (0.016)∗∗∗
0.113
0.134
−0.162
−0.182
(0.010)∗∗∗ (0.007)∗∗∗
(0.010)∗∗∗ (0.007)∗∗∗
−218.770
−262.035
311.177
354.273
(19.333)∗∗∗ (13.258)∗∗∗ (19.290)∗∗∗ (13.713)∗∗∗
0.521
0.475
0.521
0.475
(continued)
The Criminal Careers of Italian Mafia Members
257
Table 4 (continued)
Observations
Base outcome is the average recruitment age (28–
39 years old)
Early recruits (<28 years
Late recruits (>39 years
old)
old)
3396
5972
3396
5972
Notes: standard errors reported in parenthesis. “MO” refers to the first mafia offence. ∗, ∗∗ and ∗∗∗
indicate statistical significance at the 5, 1 and 0.1 per cent level, respectively. Early recruits are
offenders whose age at the entrance into the mafia organization is lower than the 25th percentile of
the distribution of the age at entrance into the mafia organization, while late recruits’ age at
entrance into the mafia is higher than the 75th percentile. Base outcome for the mafia association
is the Apulian mafia
Early recruits tend to have completed less years of education compared to offenders who joined the mafia at an “average” age, while late recruits are on average more
educated. This result is robust to the two different specifications reported in Table 4
and it is not driven by a temporal trend to recruit younger (and more educated)
offenders, as all models control for the offender’s year of birth. The positive relationship between years of education and age at first mafia offense might have different explanations. More educated offenders may be better able to hide their criminal
behavior to law enforcement authorities, which did not detect some mafia association crimes they committed while they were younger. It might also be that offenders
with higher education face more career options, and thus commit to the involvement
into the mafia organization at a later stage. Given that a higher education is associated with a late entrance into the mafia organization, the educational background
appears to be a factor influencing the recruitment into the mafia, as suggested by
other studies on organized crime (Albini 1971; Carvalho and Soares 2016; Ciconte
1992; Hess 1993; Sales 2015). This literature indicates that illiterate people from
lower socio-economic background are more likely to join organized criminal groups
(see Savona et al. 2017b). Official reports of Italian anti-mafia institutions also cite
the relation between poor education, low economic conditions and scarce access to
legitimate jobs as risk factors for youths’ involvement in the mafias (Commissione
Parlamentare Antimafia 2012; Direzione Investigativa Antimafia 2002).
Two specific criminal features characterize early recruits compared to offenders
who joined the mafia at an “average” age: early recruits were on average more versatile in the type of crimes committed before entering the mafia organization, and
they committed on average less serious offences. Conversely, compared to offenders
who joined the mafia organization at an “average” age, late recruits committed a
lower number of violent crimes before entering the mafia and they were less versatile offenders. Hence, the propensity to commit different types of offences appear to
be a relevant criminal feature influencing the timing of entrance into the mafia
organization.
Compared to the baseline group, early and late recruits are characterized by
sharp differences in the type of crimes committed before joining the mafia organization. Offenders who, before joining the mafia, committed more robbery, theft and
258
E. U. Savona et al.
related crimes, more arson and assault crimes, more administrative and financial
crimes and more associative crimes13 tend to enter the mafia organization at younger
age. Conversely, offenders committing (before the first mafia association crime)
more drug-related crimes and more white-collar crimes join the mafia organization
on average at a later age.
Turning to the spatial dimension of mafia involvement, available data highlight
that Italian mafia organizations are rooted in specific Italian provinces and regions.
The province of birth is a strong predictor of the type of mafia organization that
offenders will join: for example, of 417 offenders born in the province of Caserta (in
the Campania region) and for whom the mafia association is known, 99.52%
(n = 415) belong to the Camorra.14 Moreover, the great majority of offenders
(84.45%) reside in the region of birth, thus suggesting that most offenders are
recruited in the region where they were born. Nonetheless, some mafia organizations appear to be more prone to relocate their criminal activities outside the region
where they are traditionally rooted. Compared to other mafia members, ‘Ndrangheta
members are significantly more likely to reside in a region different from the one
where they were born, while the opposite is true for Sicilian mafia members.15
The micro level analysis of mafia members’ criminal careers highlighted that
violent and versatile offenders with less years of education tend to enter the mafia
organization when they are younger. Offenders are usually recruited in the province
or region of birth. These findings offer valuable insights for designing specific crime
prevention strategies tailored to individuals who are more likely to be recruited into
the mafia.
As a complementary although separate exploration, the study explored the differences in the criminal career of mafia members before and after the recruitment
into the mafias. The moment of recruitment was operationalised as the year of the
first mafia association offence in each individual’s criminal career. To this aim, the
analysis focused solely on the subset of individuals with sufficient and reliable
information on the mafia association offenses (n = 5993).16
As a preliminary analysis the dimensions of the criminal career (duration, frequency, specialization, escalation, number of crimes, and average seriousness) were
computed for both periods before and after recruitment (Table 5).17 The statistical
“associative crimes” excludes the mafia association crimes, thus grouping together the crimes of
“criminal association” and “drug trafficking criminal association”.
14
These statistics refer to the mafia association variable before data imputation, see Section 6 for
more details.
15
These considerations arise from logistic regression models having as dependent variable the
offender’s mafia association (Savona et al. 2017a, p. 227).
16
The initial 11,144 mafia members were filtered based on the number of crimes they committed,
maintaining in the sample only those members who committed at least four crimes. For each member was then computed the probability that crimes with no year were the first or last crimes of the
member’s career. These two passages lowered the sample size from 11,144 to 5993.
17
To test the robustness of the choice of the year of the first mafias offence as a proxy for the
recruitment year, we calculated the distributions of the dimensions moving the recruitment moment
backwards in time of 1, 3 and 5 years. The analyses focused on those members with information
13
The Criminal Careers of Italian Mafia Members
259
Table 5 Mean values of the criminal career’s parameters of mafia members before and after
recruitment
Number of
Duration Frequency Specialization Escalation crimes
Before 11.74
1.40
0.67
6.36
10.46
After
9.50
1.74
0.64
−5.39
7.74
Average
seriousness
57.27
82.70
significance of the differences before and after recruitment was assessed with
T-tests, showing that escalation is significantly lower after recruitment and the average seriousness is significantly higher. This suggests an escalation in the criminal
pattern of mafia members before recruitment, followed by more serious offenses
after the recruitment moment.18
The average seriousness shows an almost constant escalation in the years before
recruitment, followed by an almost symmetric de-escalation (Fig. 4). A very similar
situation is found by the average number of crimes (Fig. 5). These trends suggest
that in the years before commission of the first mafia offence, the individuals in the
sample are already engaged in criminal activity, with a slight increase in the years
immediately before the mafia offence. Yet the offences’ seriousness is increasingly
higher, suggesting that on average the individuals in the sample may go through a
pattern of “criminal training” before entering the mafias. Surprisingly, after entrance
in the mafias, the criminal activity de-escalates both in terms of number of offences
and seriousness. Nevertheless, this second pattern could be biased due to the limitations of the data. For example, it is likely that investigations for mafia offences
entail pretrial detention or conviction sentences. These prison time for mafia association and other serious crimes likely account for an important part of the de-­
escalation after recruitment. Prison time certainly reduces the offending capacity of
the mafia members, which in turn depresses the average values of the population.
While the data lack detailed information on the actual detention periods, the criminal career of a given mafia member includes both years with committed crimes
(active years), and years with no criminal records (silent years). To analyze the
effect of the silent years on the results, a more specific test focused on the active
years only. Starting from the year of recruitment t0, each step backwards/forwards
represents the previous/following year in which the mafia member was “active”, i.e.
committed at least one crime. Contrary to the analysis across time, which comprises
periods of both criminal activity and inactivity, the analysis across active years only
accounts for periods of criminal activity.
Indeed, the after-recruitment behaviour clearly differs when analysing the seriousness over active years (Fig. 6). After increasing before recruitment and peaking
about all the six different dimensions, and on the four time-buffers defined for 0, 1, 3 and 5 years.
The correlation was computed, for each dimension, before and after recruitment, and across the
time-buffers. Results show strong correlations on average, pointing out the robustness of the
choice. Further details about the robustness checks are in (Savona et al. 2017a).
18
The offence of mafia association is excluded from the calculation of the seriousness, as it would
have increased the average seriousness score for all mafia members.
260
E. U. Savona et al.
Fig. 4 Average seriousness before and after recruitment, across both active and silent years. t0
represents the year of recruitment
Fig. 5 Average number of crimes before and after recruitment, across both active and silent years.
t0 represents the year of recruitment
in the year of recruitment, the average crime seriousness stabilises. Removing periods of inactivity (likely to include prison time) uncovers that the seriousness of
crimes after recruitment maintains higher levels than before recruitment. Similarly,
the analysis of the average number of crimes across active years confirms the escalation before recruitment, but points up a slight increasing number of crimes after
recruitment (Fig. 7).
Given the establishment of a high level of seriousness after the recruitment
moment, the next issue to be addressed was how the number of crimes changed
(before and after recruitment) across crime categories. Descriptive statistics show
that recruitment into mafias induce a drop in some low-seriousness categories such
The Criminal Careers of Italian Mafia Members
261
Fig. 6 Average seriousness before and after recruitment, across active years. t0 represents the year
of recruitment
Fig. 7 Average number of crimes before and after recruitment, across active years. t0 represents
the year of recruitment
as financial offences, robbery and theft, overcompensated by (i) the appearance of
mafia association offenses, and (ii) a rise in high-seriousness categories such as drug
trafficking and criminal association, and extortion (Fig. 8).
Overall, the analysis of the seriousness and number of crimes before and after
recruitment shows a remarkable escalation. Mafia members increase their offending
frequency and particularly the seriousness of the offenses. The path to mafia recruitment suggests that, on average, candidates need to show their criminal skills to be
admitted in the organization. Recruitment impacts significantly on mafia members’
offending patterns. The number of crimes, when considering only active years, continues to escalate, while the seriousness stabilises at high levels. Recruitment into
262
E. U. Savona et al.
Fig. 8 Average number of crimes (per ID) per crime category, before and after recruitment into
mafias
The Criminal Careers of Italian Mafia Members
263
the mafias seems therefore to perform a catalyst function in the criminal career of
the members, expanding offending opportunities both in number and seriousness.
The general belief that sees recruitment into mafias as a starting point in the criminal career of a mafia member, is therefore challenged. In particular, the data present
a new alternative and interesting point of view, which finds recruitment into mafias
as a secure settlement point.
Discussion and Conclusions
The results of the analysis at Micro level shed light on the entire study helping to
discuss its contribution to the knowledge of the criminal trajectories of Italian
Mafiosi.
The first point to be discussed is how this study in general combines with the
theoretical framework designed by Blumstein and colleagues (Blumstein et al.
1988; Piquero et al. 2003). Having chosen the same parameters identified by
Blumstein, the question could be not relevant. On the contrary, considering that five
of the six parameters have been operationalized ex post, the question becomes relevant because the data collected and used for the operationalization process have
influenced the development of the study. The high level of missing information in
the PMM dataset for socio-demographic variables should be considered as a general
limitation to the interpretation of its results (Table 1).
Having considered this limitation and taken into account those parts of the study
that rely on robust data, the answer to the first point is positive. This means that the
results of this study are consistent with the general framework on criminal careers
and with other studies related to the criminal careers of organized crime members
(Kleemans and van Koppen 2014; Van Koppen et al. 2010; Francis et al. 2013;
Fuller et al. 2019). What results is that Italian “Mafiosi” are not different in their
criminal trajectories from other serious criminals as pointed out in section “The
Criminal Trajectories of mafiosi”. Results presented in the other sections, exclude
that Italian mafia members who have been arrested, present relevant differences
with other serious criminals in the five parameters considered, confirming the validity of the conceptual framework. Considering that this study is the first to analyse
the criminal trajectories of mafia members this conclusion is relevant. Same homogeneity is found when in the meso analysis (section “Similarities and Differences of
Italian mafiosi Across Their Criminal Organizations”) convicted criminals are analyzed across the four criminal organizations. Marginal differences are observed in
the distribution of the five parameters allowing to conclude that the seriousness of
the criminal behavior of mafia members is prevalent on the existing differences
among criminal organizations. It should be considered here that even if the variable
“mafia Association” was present in the database with relevant missing data, this
study has overcome this obstacle by creating a new variable (mafia association
aggregated) following the procedure explained in footnote n. 6.
The second point is the relation of the results of this study with the main project
on the recruitment of members from criminal organizations this study is part of.
264
E. U. Savona et al.
Results from the micro-analysis (section “When, Why and Where Do mafiosi Enter
in Their Criminal Organization?”) help to draw the picture in relation to the Italian
criminal organizations. The analysis shows that early recruited members tend to
have less education compared to offenders who joined the mafia at an “average”
age; conversely, late recruits are on average more educated. Moving from the age of
recruitment to the criminal records of offenders, that are the most robust data of the
database, it emerges that early recruited offenders have committed a variety of common crimes, whereas late recruited offenders have committed more drug-related
crimes and white collar crimes. Filtering these results with the impact of the five
parameters, it appears that late recruiters with a better education are more specialized in mafia crimes than younger recruiters. Late recruitment, better education and
crime specialization are the three components that drive the different recruitment
steps in organized crime. There is a necessary caveat: smart offenders with a better
education and being more aware of the risks related to mafia crimes could have
avoided the conviction and be not present in the database considered.
These results have relevant policy implications for project PROTON because
they can help in driving policies for curbing recruitment processes from criminal
organizations. If early recruited are young, less educated, committing a variety of
crimes and late recruited are more educated and more specialized in crimes committed, policies for preventing these two targets need to be different. It could be assumed
that prevention policies of primary (family level) and secondary (school level)
socialization could work effectively in driving young people to legitimate activities
making the illegitimate ones less attractive. On the contrary, criminal policies with
severe sanctions and sophisticated investigations could be more effective in deterring the recruitment of older and more educated offenders that are specialized in
mafia crimes.
Targets and policies are consistent with the recent changes of Italian organized
crime: declining in participation, as the results of this study show, but becoming
more sophisticated in its infiltration in the legitimate businesses. The 2018
Parliamentary Antimafia Commission writes in its report that there is a common
trend in all the territories and organizations consisting in the progressive reduction
of violence in favor of exchange relationships in the illegal and legal markets (CPA
2018, p. 13). These are the relationships that request more sophisticated and well-­
educated personnel. This evolution that could be generalized for all the Italian criminal organizations is anticipated by recent investigations of mafia prosecutors. In the
ad-hoc meeting on Proton project held in Milan on September 15, 2018, the general
prosecutor of Naples declared that from recent investigations it becomes clear that
the recent heads of Camorra organizations are legitimate entrepreneurs.
Entrepreneurs have replaced old “traditional” bosses possibly because they are better able to manage criminal organization in a new environment where the exchange
between the illegal and legal markets are particularly intense.
This study has analyzed the criminal trajectories of Italian “Mafiosi” providing
relevant data for discussing the characteristics of those entering in the Italian criminal organizations. The results, despite some relevant missing data, have opened a
window on the understanding of the mechanisms related to the recruitment processes, introducing policy scenarios that are the object of other parts of Proton proj-
The Criminal Careers of Italian Mafia Members
265
ect. A more complete view to understand today’s processes could be possible
combining the 30 years transformations of Italian organized crime with the policies
used for combatting it. Data used by this project are a good start but additional data
are needed, especially on those socio-demographic variables that have relevant
missing values in this study. These data combined with criminal trajectories of
Italian Mafiosi here analyzed, could produce more focused lessons for future policies against criminal organizations in Italy and beyond.
As explained in the introduction, this chapter develops a statistical analysis of the
data contained in the dataset, following Blumstein’s framework. Its conclusions are
in the comments to the different analyses conducted. The analysis of these data in
relations to the transformations of Italian mafias is postponed to another publication, part of project PROTON. This analysis will start using the Blumstein patterns
to understand and explain the transformations of mafias in Italy in the last 30 years.
Linking these transformations with past and present antimafia policies could help in
developing instruments and offering suggestions to mitigate and reduce the expansion of mafias, and consequently their attractiveness and potential for recruitment.
References
Albini, J. L. (1971). The American Mafia: Genesis of a legend. New York: Appleton-Century-Crofts.
Antimafia, D. I. (2002). Relazione del Ministro dell’Interno al Parlamento sull’attività svolta e sui
risultati conseguiti dalla Direzione Investigativa Antimafia. 1° Semestre 2002. In Rome.
Berlusconi, G. (2014). Italian Mafia. In G. Bruinsma & D. Weisburd (Eds.), Encyclopedia of criminology and criminal justice. New York: Springer.
Blokland, A. a. J., & Nieuwbeerta, P. (2005). The effects of life circumstances on longitudinal trajectories of offending∗. Criminology, 43(4), 1203–1240. https://doi.
org/10.1111/j.1745-9125.2005.00037.x.
Blumstein, A., Cohen, J., Roth, J. A., & Visher, C. A. (1986). Criminal careers and ‘career criminals’ (National Research Council (U S. ) panel on research on criminal careers). Washington,
DC: National Academy Press.
Blumstein, A., Cohen, J., & Farrington, D. P. (1988). Criminal career research: Its value for criminology. Criminology, 26(1), 1–35. https://doi.org/10.1111/j.1745-9125.1988.tb00829.x.
Carvalho, L. S., & Soares, R. R. (2016). Living on the edge: Youth entry, career and exit in
drug-selling gangs. Journal of Economic Behavior & Organization, 121, 77–98. https://doi.
org/10.1016/j.jebo.2015.10.018.
Ciconte, E. (1992). ‘Ndrangheta dall’unità a oggi. Bari: Editori Laterza.
Ciconte, E. (2014). Origins and development of the ‘Ndrangheta. In N. Serenata (Ed.), The
‘Ndrangheta and Sacra Corona Unita (Vol. 12, pp. 33–50). Cham: Springer International
Publishing. https://doi.org/10.1007/978-3-319-04930-4_3.
Commissione Parlamentare Antimafia. (2012). Relazione sulla prima fase dei lavori della
Commissione, con particolare riguardo al condizionamento delle mafie sull’economia, sulla
societa` e sulle istituzioni del Mezzogiorno. Commissione Parlamentare d’Inchiesta sul
Fenomeno della Mafia e sulle altre Associazioni Criminali, anche Straniere, Doc. XXIII, n. 9,
XVI legislatura. Rome.
Commissione Parlamentare Antimafia. (2018). Relazione conclusiva. Rome: Commissione parlamentare di inchiesta sul fenomenodella mafia e sulle altre associazioni criminali, anche straniere. Doc. XXIII, n. 38, Rome.
266
E. U. Savona et al.
Europol. (2013). Threat assessment: Italian organized crime. In June.
Farrington, D. P. (1992). Criminal career research in the United Kingdom. The British Journal of
Criminology, 32(4), 521–536.
Farrington, D. P., MacLeod, J. F., & Piquero, A. R. (2016). Mathematical models of criminal careers. Journal of Research in Crime and Delinquency, 53(3), 336–355. https://doi.
org/10.1177/0022427815620237.
Francis, B., Humphreys, L., Kirby, S., & Soothill, K. (2013). Understanding criminal careers
in organised crime (Report). Home Office. http://eprints.lancs.ac.uk/67009/. Accessed 19 Oct
2017.
Fuller, G., Morgan, A., & Brown, R. (2019). Criminal histories of Australian organised crime
offenders (no. 567). Canberra: Australian Institute of Criminology.
Glueck, S., & Glueck, E. (1968). Delinquents and nondelinquents in perspective. Cambridge:
Harvard Univ Press. Accessed 20 Jan 2018.
Hess, H. (1993). Mafia: Le origini e la struttura. Roma-Bari: Editori Laterza.
Jones, B. L., & Nagin, D. S. (2013). A note on a Stata plugin for estimating group-based
trajectory models. Sociological Methods & Research, 42(4), 608–613. https://doi.
org/10.1177/0049124113503141.
Kleemans, E. R., & de Poot, C. J. (2008). Criminal careers in organized crime and social
opportunity structure. European Journal of Criminology, 5(1), 69–98. https://doi.
org/10.1177/1477370807084225.
Kleemans, E. R., & van Koppen, M. V. (2014). Careers in organized crime. In D. Weisburd &
G. Bruinsma (Eds.), Encyclopedia of criminology and criminal justice. New York: Springer.
Kyvsgaard, B. (2002). The criminal career: The Danish longitudinal study. Cambridge: Cambridge
University Press.
Laub, J. H., Nagin, D. S., & Sampson, R. J. (1998). Trajectories of change in criminal offending:
Good marriages and the desistance process. American Sociological Review, 63(2), 225–238.
https://doi.org/10.2307/2657324.
Lussier, P., Tzoumakis, S., Cale, J., & Amirault, J. (2010). Criminal trajectories of adult sex offenders and the age effect: Examining the dynamic aspect of offending in adulthood. International
Criminal Justice Review, 20(2), 147–168. https://doi.org/10.1177/1057567710368360.
Massari, M. (2014). The Sacra Corona Unita: Origins, characteristics, and strategies. In
The ‘Ndrangheta and Sacra Corona Unita (pp. 101–116). Cham: Springer. https://doi.
org/10.1007/978-3-319-04930-4_7.
Nagin, D. S. (2016). Group-based trajectory modeling and criminal career research.
Journal of Research in Crime and Delinquency, 53(3), 356–371. https://doi.
org/10.1177/0022427815611710.
Paoli, L. (2008). The decline of the Italian mafia. In Organized crime: Culture, markets and policies (pp. 15–28). New York: Springer. https://doi.org/10.1007/978-0-387-74733-0_2.
Paoli, L. (2014). The Italian mafia. In L. Paoli (Ed.), The Oxford handbook of organized crime
(pp. 121–141). New York: Oxford University Press.
Piquero, A. R., Farrington, D. P., & Blumstein, A. (2003). The criminal career paradigm. Crime
and Justice, 30, 359–506.
Piquero, A. R., Raymond, P., Paul, M., Robert, B., & Charles W. D. (1999). Onset Age and Offense
Specialization. Journal of Research in Crime and Delinquency, 36(3), 275–299. https://doi.org
/10.1177/0022427899036003002.
Rhodes, W. (1989). The Criminal Career: Estimates of the Duration and Frequency of Crime
Commission. Journal of Quantitative Criminology, 5(1), 3–32. https://doi.org/10.1007/
BF01066259.
Sales, I. (2015). Storia dell’Italia mafiosa: perché le mafie hanno avuto successo. Soveria
Mannelli: Rubbettino.
Savona, E. U., Calderoni, F., Campedelli, G. M., Comunale, T., Ferrarini, M., & Meneghini, C.
(2017a). Recruitment into mafias: criminal careers of mafia members and mafia bosses (report
The Criminal Careers of Italian Mafia Members
267
on factors relating to OC) (pp. 192–312). Milan: UCSC-Transcrime. https://www.projectproton.eu/wp-content/uploads/2018/01/D1.1-Report-on-factors-relating-to-OC.pdf.
Savona, E. U., Calderoni, F., Superchi, E., Comunale, T., Campedelli, G. M., Marchesi, M. E.,
& Kamprad, A. (2017b). Systematic review of the social, psychological and economic factors
relating to criminalisation and recruitment to OC (report on factors relating to OC) (pp. 5–71).
Milan: UCSC-Transcrime. https://www.projectproton.eu/wp-content/uploads/2018/01/D1.1Report-on-factors-relating-to-OC.pdf.
Serenata, N. (2014). Introduction. In N. Serenata (Ed.), The ‘Ndrangheta and Sacra
Corona Unita (Vol. 12, pp. 1–13). Cham: Springer International Publishing. https://doi.
org/10.1007/978-3-319-04930-4_1.
van de Rakt, M., Murray, J., & Nieuwbeerta, P. (2012). The long-term effects of paternal imprisonment on criminal trajectories of children. Journal of Research in Crime and Delinquency,
49(1), 81–108. https://doi.org/10.1177/0022427810393018.
Van Koppen, M. V., de Poot, C. J., Kleemans, E. R., & Nieuwbeerta, P. (2010). Criminal trajectories in organized crime. The British Journal of Criminology, 50(1), 102–123. https://doi.
org/10.1093/bjc/azp067.
van Onna, J. H. R., van der Geest, V. R., Huisman, W., & Denkers, A. J. M. (2014). Criminal
trajectories of white-collar offenders. Journal of Research in Crime and Delinquency, 51(6),
759–784. https://doi.org/10.1177/0022427814531489.
Investigating the Psychological Profile
of Organized Crime Members
Gerardo Salvato, Maria Laura Fiorina, Daniela Ovadia, Gabriele De Maio,
Elisa Francescon, and Gabriella Bottini
Introduction
Criminal behaviour is a growing worldwide public issue. Likely other human
behaviours, it arises from a convergence of different factors. Many studies have
highlighted the importance of the influence of neurobiological, genetics and socio-­
cultural factors in criminal behaviours (Glenn and Raine 2014; Salvato et al. 2014).
Psychology and neuroscience gained much attention in forensic settings, and its
scientific instruments may help in shedding light on the neurobiological features
characterising complex behaviours.
Over the last decades, researchers have tried to explore and understand the possible
link between criminal behaviour and personality traits. Personality is a multifactorial
construct and refers to individual patterns of thinking, feeling and behaving. The
majority of psychological studies on criminal behaviour provide evidence concerning
specific classes of perpetrators or more in general, in non-organized crime (NOC)
offenders. The available literature on the topic reveals that among the classifications
present in the Diagnostic and Statistical Manual of Mental Disorders (DSM)
(American Psychiatric Association 2013), the most prevalent in NOC criminals are
narcissist, paranoid, borderline, and antisocial personality disorders (Hart et al. 1993;
DeRuiter and Greeven 2000; Blackburn et al. 2003; Gunter et al. 2008; Sansone and
Sansone 2009; Falk et al. 2017; Shepherd et al. 2018). Specifically, the antisocial
G. Salvato · G. Bottini (*)
Department of Brain and Behavioral Sciences, University of Pavia, Pavia, Italy
Cognitive Neuropsychology Centre, ASST “Grande Ospedale Metropolitano” Niguarda,
Milan, Italy
NeuroMi, Milan Centre for Neuroscience, Milan, Italy
e-mail: g.bottini@unipv.it
M. L. Fiorina · D. Ovadia · G. De Maio · E. Francescon
Department of Brain and Behavioral Sciences, University of Pavia, Pavia, Italy
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_11
269
270
G. Salvato et al.
personality disorder is a common psychological condition in criminals that correlates
with aggressive behaviour (DeRuiter and Greeven 2000; Blair 2001; Gunter et al.
2008; Falk et al. 2017). According to the DSM-5, it is characterised by inflexible,
maladaptive, and persistent personality traits causing significant functional impairment or subjective distress. Antisocial offenders initiate their criminal activities from
an earlier age than others and are more likely to express violent behaviours.
Interestingly, the analysis of the scientific literature reveals that only a few psychological studies have focused on prisoners involved in OC (Schimmenti et al.
2014; Craparo et al. 2018). One of the reasons could be that several penitentiary
systems foresee particularly restrictive imprisonment conditions for individuals
belonging to OC, which prevents any contacts with third parties (including researchers). Furthermore, these subjects are often very reluctant to any form of cooperation, including the involvement in psychological studies.
There is only one available study specifically assessing personality disorders in
OC prisoners (Craparo et al. 2018). Craparo et al. (2018) have investigated the presence of personality disorders in a sample of 20 members belonging to the Italian
criminal organisations Cosa Nostra (ten subjects) and Camorra (ten subjects). They
have administered prisoners with the Millon Clinical Multiaxial Inventory-III
(MCMI-III; (Millon and Meagher 2004)), and other tests, comparing the results
between the two groups. They have shown that there were no clinically significant
differences between the two groups at the MCMI-III except for the Masochism
personality scores, which were higher in Camorra members.
Although interesting, this study has some limitations specifically concerning: (i)
the very small sample size, and (ii) the absence of a control group of NOC criminals. In this scenario, in the current study, we aimed at investigating personality
disorders in a larger sample of OC offenders compared to a NOC prisoners group,
to reveal distinctive personality traits of OC members. To this aim, we administered
100 prisoners (50 OC and 50 NOC) with a psychological test that identifies the presence of specific personality disorders. Based on previous findings we expect to find
masochistic and/or antisocial personality disorders in OC criminals.
Materials and Methods
Participants
One hundred prisoners were randomly recruited from two Italian Prisons. According
to their legal records, we assessed 50 OC (age: M = 48.7, SD = 10.2; education:
M = 10.7, SD = 3) and 50 NOC (age: M = 46.6, SD = 13.4; education: M = 10.9,
SD = 3.3) male prisoners. Participants were matched in age and years of education.
All participants were native Italian speakers and had normal or corrected-to-normal
vision. They had no history of neurologic or psychiatric disorders. Informed consent
was obtained prior to participation in the study according to the Declaration of
Helsinki. The Italian Department of Prison Administration and Department of Brain
and Behavioural Sciences of the University of Pavia (protocol n°010) approved the
Investigating the Psychological Profile of Organized Crime Members
271
study. All participants read and signed an informed consent form in which the aims
of the study were explained.
Tests
Firstly, participants were administered with two neuropsychological tests assessing
the integrity of the global cognitive functioning and intelligence: the Addenbrooke’s
Cognitive Examination-Revised (ACE-R) (Mioshi et al. 2006; Siciliano et al. 2016),
and the Raven’s Coloured Progressive Matrices (CPM) (Raven 1958).
Following the neuropsychological screening, they completed the Millon Clinical
Multiaxial Inventory - Third Edition (MCMI-III) (Millon and Meagher 2004). The
MCMI-III is a psychological assessment tool intended to provide information on
personality traits and psychopathology, including specific disorders outlined in the
DSM. It consists of 175 true-false items. It distinguishes 11 moderate personality
disorder scales: schizoid, avoidant, depressive, dependent, histrionic, narcissistic,
antisocial-sadistic, compulsive, negativistic, masochistic, and 3 severe personality
pathology scales: schizotypal, borderline, paranoid.
Statistical Analyses
Data were analysed using SPSS 20 (Statistical Package for Social Science, Chicago,
Illinois). As the first step, we evaluated the two screening tests to exclude participants
with neuropsychological deficits that could influence the personality traits assessment.
According to the tests Italian validations (Basso et al. 1987; Siciliano et al. 2016), none
of the participants presented with scores below the cut-offs in the global cognitive
functioning and intelligence tests. Then, we evaluated personality traits and disorders
of each participant using thresholds according to the Italian validation of the MCMIIII. Accordingly, a cut-off score ≥75 for each of the personality disorder scales was
used to be indicative of a probable diagnosis. For each participant, scale scores were
binary classified as 0 disorder absent (score <75) or 1 disorder present (score ≥75).
Then, we used a binary logistic regression to predict prisoners membership (OC, NOC)
from a set of categorical predictor variables resulting from each dimension explored
with the MCMI-III for moderate and severe personality disorder scales.
Results
We found no difference between the OC and NOC prisoners in terms of age
(F(1,99) = 0.8; p = 0.382), and education (F(1,99) = 0.1; p = 0.706). We did not exclude
any participants from subsequent analyses as they showed a normal global cognitive
functioning and intelligence level.
272
G. Salvato et al.
A logistic regression analysis was performed with group (OC, NOC) as the
binary dependent variable and the scales of moderate personality disorders (schizoid, avoidant, depressive, dependent, histrionic, narcissistic, antisocial-sadistic,
compulsive, negativistic, masochistic) as predictors. The years spent in prison were
included in the model as constant. This final model was significant (omnibus test
χ2(2) = 32.6; p < 0.001; Hosmer and Lemeshow χ2(8) = 8.7; p = 0.36)), and it explained
37.2% of Nagelkerke’s pseudo-variance. We found that holding the years of detention as constant, the odds of an OC prisoner being diagnosed with masochistic personality disorder were 0.25 times greater than the odds for a NOC prisoner (Wald
χ2(1) = 3.8; p = 0.049; OR = 0.25; CI = 0.06–0.99); the other scales were removed
from the model (all ps > 0.05). A second logistic regression analysis performed
using the three severe personality disorder scales (schizotypal, borderline, paranoid)
as predictors of membership did not reach the statistical significance.
As Craparo et al. (2018) have shown that masochistic personality traits were
more likely to occur within a small group of Cosa Nostra compared to Camorra
members, we checked for possible influence of the geographical origin region/
OC-type organisations of our OC members showing a masochistic personality disorder. We performed a chi-square non-parametric test on geographical origin region/
OC-type organisations of the OC prisoners and results did not show any statistical
prevalence of a geographical region/OC-type organisations (x2(3) = 1.2; p = 0.753).
Discussion
Understanding the personality profile of OC members is a crucial factor that may
help in identified tailored rehabilitative treatments and policymakers in devising
effective prevention policies. In the current study, we explore the personality traits
specifically distinguishing the OC from NOC prisoners. Results showed that masochistic personality disorder was more likely to occur in OC than NOC prisoners.
Our findings expand and substantiate very recent evidence, in which masochistic
personality traits have been found in a small group (ten participants) of the Cosa
Nostra members (Craparo et al. 2018). In our case, we did not find in the OC group
presenting with a masochistic personality disorder any influence due to the specific
prevalence of Mafia-type organisations or prisoners’ region of origin in our sample.
One might speculate that the partial discrepancy between the current study and the
Craparo and colleagues’ work (2018) may be due to the very small sample of prisoners that they assessed.
From a psychological point of view, masochistic behaviours have multiple determinants. As suggested by Gabbard (2012) one of such determinant may relate to
active mastery over passively experienced trauma. Individuals who have masochistic personality disorder may recreate abusive and humiliating experiences that they
have endured in the past but from the driver’s seat point of view. Throughout this
unconscious reiteration, they avoid being ambushed by it.
Moreover, from their point of view, the relational attachment to an abusive or
sadistic person may be better than having no relationship at all (Gabbard 2012).
Investigating the Psychological Profile of Organized Crime Members
273
As suggested by Schimmenti et al. (2014), OC members have to adhere, be submitted and obedient to the willingness of their boss (Schimmenti et al. 2014). The
finding of masochist personality disorder in OC members appears to be coherent
with the framework of the organization. It is also important to note that we investigated the behavioural profile of OC members pertaining to the low-hierarchy of
the organization so that such results should be considered as selectively concerning these subjects that typically serve the organization as passive executors rather
than strategic operators.
Being obsequious and incline to sacrifice, encouraging others to take advantage
of this, is typical of the masochistic profile (Millon et al. 2009). This psychological
profile shows several ideal features for the OC that imposes hard rules and rigid
behaviours to be applied to its members. The OC members are bound by commandments based on mobster guidelines. These rules include “professional” and “private” aspects. A good example is the 5th commandment saying: “Always being
available for Cosa Nostra is a duty – even if your wife is about to give birth” indicating the imposition of absolute obedience to the organization. This kind of subjection is typical of the masochistic profile. These written guidelines have also been
defined to control and rein the rebel behaviours of the younger Mafiosi. The rite of
Mafia initiation is based on a sort of religious communion defined by obligations
and rules of correctness and solidarity and are sealed by the condition of suffering.
It is also important noticing that some caution should be paid in interpreting
results from the current study. The question arises of whether the association with
an OC group would shape the behaviour, or the individual must have some innate
personal characteristics to be associated with an OC group. Misuse and misunderstanding concerning the role of psychology and neuroscience in criminal evaluation
have been largely discussed (Gazzaniga 2011; Bottini et al. 2015; Morse 2018). A
specific psychological/behavioural characteristic does not justify and explain in an
exhaustive way the criminal behaviour or a predisposition for an individual to
become a criminal. Instead, both innate factors (genetic heritability), environment,
and the interplay between the two would contribute to the development of individual behavioural characteristics, also involving mental health and cognition (Plomin
and Asbury 2005), so that genetics and specific environmental triggers are required
for a mental disorder to develop.
To conclude, OC members may present with some specific personality traits,
which presumably arise from the interaction between innate factors and the environment. Further studies are needed to explore the neuropsychological/cognitive p­ rofile
of OC members that together with the psychological structure may contribute to
delineate an increasingly detailed cognitive-behavioural picture of OC criminals.
References
American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders
(5th ed.). Arlington: American Psychiatric Association.
Basso, A., Capitani, E., & Laiacona, M. (1987). Raven’s coloured progressive matrices: Normative
values on 305 adult normal controls. Functional Neurology, 2, 189–194.
274
G. Salvato et al.
Blackburn, R., Logan, C., Donnelly, J., & Renwick, S. (2003). Personality disorders, psychopathy
and other mental disorders: Co-morbidity among patients at English and Scottish high-security
hospitals. The Journal of Forensic Psychiatry & Psychology, 14, 111–137. https://doi.org/10.1
080/1478994031000077925.
Blair, R. J. R. (2001). Neurocognitive models of aggression, the antisocial personality disorders,
and psychopathy. Journal of Neurology, Neurosurgery, and Psychiatry, 71, 727–731. https://
doi.org/10.1136/jnnp.71.6.727.
Bottini, G., Sedda, A., & Ovadia, D. (2015). Past present and future of neuroscience and law.
Italian Journal of Criminology, 7, 17–22.
Craparo, G., David, V., Costanzo, G., & Gori, A. (2018). Cosa Nostra and the Camorra: Assessment
of personality, alexithymic traits, and attachment styles. International Journal of Law and
Psychiatry, 58, 17–26. https://doi.org/10.1016/j.ijlp.2018.02.010.
DeRuiter, C., & Greeven, P. G. J. (2000). Personality disorders in a dutch forensic psychiatric
sample: Convergence of interview and self-report measures. Journal of Personality Disorders,
14, 162–170. https://doi.org/10.1521/pedi.2000.14.2.162.
Falk, Ö., Sfendla, A., Brändström, S., et al. (2017). Personality and trait aggression profiles of
male and female prison inmates. Psychiatry Research, 250, 302–309. https://doi.org/10.1016/j.
psychres.2016.12.018.
Gabbard, G. O. (2012). Masochism as a multiply-determined phenomenon. In D. Holtzman &
N. Kulish (Eds.), The clinical problem of masochism (pp. 103–112). Plymouth: Jason Aronson.
Gazzaniga, M. S. (2011). Neuroscience in the courtroom. Scientific American, 304, 54–59. https://
doi.org/10.1038/scientificamerican0411-54.
Glenn, A. L., & Raine, A. (2014). Neurocriminology: Implications for the punishment, prediction
and prevention of criminal behaviour. Nature Reviews. Neuroscience, 15, 54–63. https://doi.
org/10.1038/nrn3640.
Gunter, T. D., Arndt, S., Wenman, G., et al. (2008). Frequency of mental and addictive disorders
among 320 men and women entering the Iowa prison system: use of the MINI-plus. Journal of
the American Academy of Psychiatry and the Law, 36, 27–34. doi: 36/1/27 [pii].
Hart, S. D., Dutton, D. G., & Newlove, T. (1993). The prevalence of personality disorder among
wife assaulters. Journal of Personality Disorders, 7, 329–341. https://doi.org/10.1521/
pedi.1993.7.4.329.
Millon, T., & Meagher, S. E. (2004). The Millon Clinical Multiaxial Inventory-III (MCMI-III).
In Comprehensive handbook of psychological assessment (Personality assessment) (Vol. 2,
pp. 109–121). Hoboken: Wiley.
Millon, T., Millon, C., Davis, R., & Grossman, S. (2009). MCMI-III Manual (4th ed.). Minneapolis,
MN: Pearson Education, Inc.
Mioshi, E., Dawson, K., Mitchell, J., et al. (2006). The Addenbrooke’s Cognitive Examination
revised (ACE-R): A brief cognitive test battery for dementia screening. International Journal
of Geriatric Psychiatry, 21, 1078–1085. https://doi.org/10.1002/gps.1610.
Morse S. J. (2018). Neuroscience in Forensic Contexts: Ethical Concerns. In E. Ezra & H. Griffith
(Eds.), Ethics Challenges in Forensic Psychiatry and Psychology Practice (pp. 132–158). U of
Penn Law School, Public Law Research Paper No. 17-8. Available at SSRN: https://ssrn.com/
abstract=2923354
Plomin, R., & Asbury, K. (2005). Nature and nurture: Genetic and environmental influences on
behavior. The Annals of the American Academy of Political and Social Science, 600, 86–98.
https://doi.org/10.1177/0002716205277184.
Raven, J. C. (1958). Guide to using the Coloured Progressive Matrices. H. K. Lewis & Co. https://
psycnet.apa.org/record/1960-00085-000.
Salvato, G., Dings, R., & Reuter, L. (2014). Culture, neuroscience and law. Frontiers in Psychology,
5. https://doi.org/10.3389/fpsyg.2014.01196.
Sansone, R. A., & Sansone, L. A. (2009). Borderline personality and criminality. Psychiatry
(Edgmont), 6(10), 16.
Investigating the Psychological Profile of Organized Crime Members
275
Schimmenti, A., Caprì, C., La Barbera, D., & Caretti, V. (2014). Mafia and psychopathy. Criminal
Behaviour and Mental Health, 24, 321–331. https://doi.org/10.1002/cbm.1902.
Shepherd, S. M., Campbell, R. E., & Ogloff, J. R. P. (2018). Psychopathy, antisocial personality
disorder, and reconviction in an Australian sample of forensic patients. International Journal
of Offender Therapy and Comparative Criminology, 62, 609–628. https://doi.org/10.1177/03
06624X16653193.
Siciliano, M., Raimo, S., Tufano, D., et al. (2016). The Addenbrooke’s Cognitive Examination
Revised (ACE-R) and its sub-scores: Normative values in an Italian population sample.
Neurological Sciences, 37, 385–392. https://doi.org/10.1007/s10072-015-2410-z.
Traffickers in Transit: Analysing
the Logistics and Involvement Mechanisms
of Organised Crime at Logistical Nodes
in the Netherlands: Empirical Results
of the Dutch Organised Crime Monitor
Renushka Madarie and Edwin W. Kruisbergen
Introduction
As demonstrated by the analysis of mobile bandits in the eighteenth century (Blok
2001) or Italian mafia families that evolved in the nineteenth century (if not earlier;
Paoli 2004), organised crime is not a new phenomenon. Nevertheless, only in the
past decades the topic of organised crime has been researched more systematically.
This research led to new insights into involvement mechanisms and criminal trajectories in organised crime. For example, Van Koppen, De Poot, Kleemans, and
Nieuwbeerta (2010) demonstrated that, unlike high volume crime offenders, organised crime offenders are generally older and have a later onset of offending (see also
Van Onna, Van der Geest, Huisman and Denkers (2014)). Researchers have also
shed light on factors that stimulate people to engage in organised crime, like deviant
group morals and financial setbacks (e.g., Kleemans and De Poot 2008; Kruisbergen
et al. 2012; Van Koppen 2013). Moreover, while positive life events, like having a
job and being married, are often considered to have a desisting effect on engagement in high volume crime, Kleemans and De Poot (2008) noted that positive life
events not always lead to desistance from engagement in organised crime. On the
contrary, family ties and jobs could in fact promote involvement in organised crime.
Jobs and family provide the social structures allowing criminals to find potential
co-offenders. Jobs could also facilitate organised crime activities if the jobs encompass a relatively high level of autonomy, mobility, or status (Kleemans and Van de
Bunt 2008).
R. Madarie
The Netherlands Institute for the Study of Crime and Law Enforcement (NSCR),
Amsterdam, The Netherlands
E. W. Kruisbergen (*)
Research and Documentation Centre (WODC), Ministry of Justice and Security,
The Hague, The Netherlands
e-mail: e.w.kruisbergen@minvenj.nl
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_12
277
278
R. Madarie and E. W. Kruisbergen
Research regarding involvement mechanisms and criminal careers in organised
crime has thus demonstrated that factors related to engagement in crime differently
affect organised crime offenders as compared to high volume crime offenders. Even
though a clearer picture of involvement mechanisms and criminal trajectories in
organised crime is now emerging, Von Lampe (2012) notes several aspects of organised crime research that deserve more scrutiny. First, Von Lampe states that most
research on transnational organised crime focuses on destination countries, whereas
the nature of criminal groups and activities in source and transit countries has been
less studied. Second, Von Lampe elaborates that in-depth analyses of the modus
operandi of criminals and the logistics of transnational organised crime activities
are the exception rather than the rule. Finally, Von Lampe explains that if coherent
theoretical frameworks are to be developed, researchers have to compare different
aspects of organised crime across different contexts (e.g., contexts defined by type
of crime). In order to do so, however, researchers have to examine clearly demarcated contexts rather than general categories that encompass a diffuse range of different types of crime.
The present study addresses each of these points by studying transnational organised crime in the Netherlands. The analyses specifically focus on one type of organised crime, namely drug trafficking, and highlight the context of one important
logistical node, namely airports. Airports are pivotal in the free flow of goods and
citizens across the world, but they are also a crucial facility for transnational organised crime. The Netherlands functions in this context as an important transit country
(Kleemans 2007). Transit crime can best be described as international illegal trade.
Although knowledge on organised crime offenders engaging in transit crime has
accumulated recently, systematic analyses of offenders operating on logistical nodes
are still rare. This chapter offers such analyses. The empirical data consists of 16
cases of the Dutch Organised Crime Monitor. Although the main analyses focus on
organised crime at airports (11 cases), organised crime at seaports (5 cases) is analysed as well in order to compare the logistics and involvement mechanisms at the
airport with the seaport.
This chapter is structured as follows: First, a brief overview of the current state
of the literature on involvement mechanisms in organised crime is presented and the
main logistical nodes in the Netherlands are described (section “Becoming Involved
in Transit Crime”). Then, the data and methods used for the analyses are outlined
(section “Data and Methods”). In the results section that follows (section “Empirical
Results”), the results of the analysis of smuggling methods employed by organised
crime offenders and the logistics of their activities are first presented. Subsequently,
the results are presented of the analysis of involvement mechanisms and the social
and occupational embeddedness of the criminal careers of the same offenders. The
chapter concludes by highlighting the main findings and discussing several dilemmas air- and seaports face in the prevention of organised crime (section “Conclusion
and Discussion”).
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
279
Becoming Involved in Transit Crime
In this section, involvement mechanisms in organised crime are first discussed. Five
general types of involvement mechanisms are explained and exemplified. Specific
job characteristics that increase the likelihood of becoming involved in organised
crime are discussed as well. The second part of this section elaborates upon the
nature of organised crime in the Netherlands as well as two settings that are highly
relevant to organised transit crime, namely the main airport in Amsterdam and the
main seaport in Rotterdam. In the third part of this section the process of international drug trafficking is described using crime script analysis.
Involvement Mechanisms in Organised Crime
The social and occupational embeddedness of organised crime provide important
clues on how people become involved in organised crime. Five established involvement mechanisms are: Social ties, work ties, deliberate recruitment, leisure activities and sidelines, and life events (Kleemans 2012). Social ties are an important
factor for becoming involved in organised crime. In the world of organised crime,
no formal mechanisms such as the law or third-party insurances exist to consolidate
trust if transactions do not work out as planned. Criminal groups operate in a hostile
and uncertain environment and, as often argued in organised crime research, therefore need other mechanisms to consolidate trust (Von Lampe and Johansen 2004).
Kleemans and Van de Bunt (1999) argue that social bonding is a solution to this
problem because of the temporal and network embeddedness. Temporal embeddedness of social ties implies that the likelihood of cheating decreases when offenders
know they will meet each other in the future and have information about each other’s past. Network embeddedness implies that group members have information
about one another and that they have a reputation to uphold in their network.
Temporal and network embeddedness of social ties also apply to work ties at the
work place. Colleagues see each other regularly and have to work together for a
longer period of time. Furthermore, colleagues form a network in which information about one another is easily shared. Colleagues therefore have a reputation to
uphold which facilitates trust in the long run. The same line of reasoning could be
applied to persons who frequently meet because they, for example, practice the
same sports, have the same hobby, or visit the same pub (Van de Bunt and Kleemans
2007). While criminal group members sometimes recruit specific persons due to
their connectedness, another reason to recruit specific persons is that they possess
specific knowledge or skills valuable to the criminal group. Such deliberate recruitment could also be aimed at persons who are in a vulnerable position, for example
due to negative life events such as financial setbacks. Negative life events at a later
age are known to increase the likelihood of people becoming involved in organised
crime (Kleemans 2012). Positive life events could also increase the likelihood of
280
R. Madarie and E. W. Kruisbergen
becoming involved in organised crime by creating more possibilities for involvement. Examples of positive life events increasing the likelihood of involvement are
getting married and becoming employed.
Legitimate jobs of organised crime offenders are often characterised by one or
more of the following factors: autonomy, mobility, and frequent social contact
(Kleemans and Van de Bunt 2008). The more autonomy a person has, the less he or
she is directly supervised, and the easier it is to engage in organised crime activities.
In a similar way, the more mobile an employee is, the harder it is to directly supervise him or her, and thus the easier it is to engage in organised crime activities. A
high level of autonomy and mobility both make it easier to conceal criminal activities by making it appear as if these activities are part of licit duties. Truck drivers
are, for example, very mobile. So when they transport drugs, they do not immediately raise suspicion because drug transports look very similar to licit transports.
Finally, jobs involving frequent social contact increase the likelihood of criminal
group members connecting to potential offenders. This is the case for, for example,
shop owners, doorkeepers, and market traders. Employees who are very mobile or
have a highly social job are also more likely to frequent so-called ‘offender convergence settings’ (Felson 2006). These settings are places where offenders meet
potential offenders and start criminal activities together, such as cafés and casinos
(Le and Gilding 2014). Offender convergence settings demonstrate that the place
where people work also affects the chance of them becoming involved in organised
crime. Because organised crime in the Netherlands often encompasses transit crime,
persons working at logistical nodes in the Netherlands could be of great interest to
organised crime groups. The most important logistical nodes in the Netherlands are
described in the next paragraph.
Transit Crime and Logistical Nodes in the Netherlands
Many types of organised crime in the Netherlands can be described as transit crime.
Transit crime mainly involves international illegal trade and uses the same opportunity structures facilitating legal economic activities (Kleemans 2007). The major
businesses of organised crime groups in the Netherlands boil down to international
smuggling activities. In these activities, the Netherlands could function as a destination country, a transit country or, in the case of cannabis and synthetic drugs, a
source country. In its function as a transit country, the Netherlands is an important
‘hub’ for the distribution of drugs to the rest of Europe. Recent reports of the
European Monitoring Centre for Drugs and Drug Addiction (EMCDDA) demonstrate that cocaine import into Europe mainly takes place through Spain and Portugal
in the south and through the Netherlands and Belgium in the north (EMCDDA
2016). The Netherlands is also an important source country for synthetic drugs such
as MDMA and amphetamines since a lot of know-how on the production is available in this country (EMCDDA 2015). The significance of the Netherlands as a
transit point, or starting point, for drug traffickers relates to its demographical,
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
281
g­ eographical and logistical characteristics (Kleemans 2014). Due to its colonial past
and government policies in the 80s and 90s, the Netherlands has a relatively large
number of inhabitants with roots in countries or regions where drugs such as
cocaine, heroin, and hash are produced on a large scale. Because of their roots, these
inhabitants are often better able to connect to these countries or regions and can
therefore be the key between foreign-based crime groups and crime groups based in
the Netherlands. In other words, their roots enable them to bridge ‘structural holes’,
i.e. to form a connection between (parts of) networks that would otherwise be poorly
connected due to geographical and social barriers (Burt 1992). For example, Turkish
persons are generally better able to connect to Turks in Turkey than Dutch persons
with no foreign roots. If a criminal group residing in the Netherlands wants to
import heroin from Turkey, a person well-connected to both the Netherlands and
Turkey is very valuable to that group. Finally, because of its geographical position,
the Netherlands is sometimes considered the gateway to Europe. Two important
logistical nodes contribute to the gateway-function of the Netherlands, namely the
main airport and seaport. These logistical nodes are discussed next.
Amsterdam Airport Schiphol
Amsterdam Airport Schiphol is the largest airport in the Netherlands and one of the
largest in Europe, with approximately 58 million passengers and 1.6 million tonnes
of airfreight being processed in 2015 (Schiphol 2017). While its enormous processing capacity and its connectedness to other parts of the world make Schiphol crucial
to the Dutch economy, these factors also make exploitation of the airport attractive
to criminal groups. Because such large volumes of persons and goods are processed,
not all passengers and airfreight can be thoroughly checked, thereby facilitating
cross-border drug transports. Several security measures have been put in place,
however, in response to large-scale drug trafficking at Schiphol. For example, in
2001, security checks of personnel working at the airport increased. Before the
implementation of security checks, personnel going from land- to airside and vice
versa could relatively easily smuggle drugs. Due to the implemented security
checks, luggage of personnel had to pass similar security procedures as luggage of
passengers. Another example concerns the so-called ‘100%-checks’ that have been
enacted since 2003 to tackle the increasing flow of drug mules smuggling cocaine
into the Netherlands (Kruisbergen 2005). All passengers and their luggage on flights
from specific countries have to pass extra security checks. Furthermore, international cooperation resulted in security measures overseas, i.e. pre-boarding checks
at Hato International Airport on the island of Curacao (part of the former Netherlands
Antilles) (Kleemans et al. 2010).1
This does not mean that every passenger is subjected to exactly the same type of checks (Van
Haaften 2013, p. 2).
1
282
R. Madarie and E. W. Kruisbergen
Port of Rotterdam
The Port of Rotterdam is one of the largest seaports in Europe. Just like the main
airport of the Netherlands, the Port of Rotterdam is an important hub for organised
crime groups (EMCDDA 2016). Between 2009 and 2012, approximately 40 tonnes
of cocaine entering the Port of Rotterdam were seized (Eventon and Bewley-Taylor
2016). In 2013, the Dutch police estimated that a quarter to half of the cocaine consumed in Europe enters via the Port of Rotterdam (Van der Ploeg 2014). Similar to
the airport, the seaport has several security systems in place aimed at detecting illicit
traffic. For example, sea containers are randomly scanned and shipping-agents have
to show certain documents before they can operate in the port.2
International Drug Trafficking: A Short Crime Script
The accumulation of knowledge on topic of drug trafficking through both scientific
research and federal investigations has made it possible to discern several key procedures in the process of drug trafficking. In so-called crime scripts, these sets of
action are schematically organised in a chronological framework (Cornish 1994). In
other words, crime scripts clarify what occurs before, during, and after a crime is
committed. Such scripts also encompass actors, equipment and locations required
for crime commission. Even though different crime groups could use different
equipment or locations, or change their modus operandi in response to police interventions, the flexibility of crime scripts allows for such improvisations. One of the
benefits of crime scripting is that it makes the different steps in the process of crime
commission easier to examine (Chiu et al. 2011). By doing so, crime scripting aids
situational crime prevention, which focuses on environmental aspects that facilitate
crime rather than on personal characteristics of perpetrators (Clarke 1997). As for
international drug trafficking, a typical crime script classified into four main activities is exemplified in Table 1 (Tompson and Chainey 2011). The content of this
crime script is based on the results of analyses of cases of the Dutch Organised
Crime Monitor (see also section “Data and Methods”).
Drug traffickers using logistical nodes such as an airport or seaport face several
challenges. For example, because of the large volume of passengers and cargo processed criminal groups have to know exactly at what time and place “their” mule,
luggage or container arrives. Furthermore, they have to make sure that their contraband passes security checks safely. To this end, they often need insider’s help.
Criminal groups could corrupt law enforcement personnel in order to evade security
A specific modus operandi employed by organised crime groups is the ‘rip-off’ (Eski and Buijt
2016). Characteristic of a rip-off is that criminal groups transport drugs to other countries by using
sea containers ordered by legitimate companies as their transport vehicle. Sea containers are transported as ordered by the legitimate companies and upon arrival, the criminal group takes out the
drug packages before they or the packages are noticed.
2
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
283
Table 1 International drug trafficking crime script
Main
activity
Preparation
Key procedures
1. Acquiring drugs
Pre-activity
2. Transporting drugs to a logistical node
Activity
3. Getting drugs on board (on the vehicle)
4. Getting drugs off the vehicle
Post-activity 5. Transporting drugs to destination
6. Stashing drugs
7. Selling drugs (sometimes after further
refinement)
8. Handling financial flows
Actors, equipment, and
location
Buying drugs from others
Producing drugs by yourself
Borrowing a car
Using acquaintances to drive
Drug mules
(Air)port employees
Customs officials
Cover materials
Borrowing a car
Using acquaintances to drive
House
Shed
Spending cash
Money laundering
checks or they could corrupt port employees to provide them with information
about the arrival of sea containers with drugs. Organised crime groups are thus
sometimes dependent on involvement of employees working at logistical nodes. But
how do these employees become involved in organised crime? The logistical bottlenecks drug traffickers have to solve (step 3 and 4 in Table 1) as well as the involvement mechanisms explaining the participation in organised crime are further
elaborated in section “Empirical Results”. First, however, the data and methods
used are discussed in section “Data and Methods”.
Data and Methods
For the present study, a selection of cases from the Dutch Organised Crime Monitor
(OCM) is used. The OCM is an ongoing research project since 1998 aimed at gaining insight into organised crime in the Netherlands. The main sources of this
research project are files of closed Dutch police investigations of criminal groups
engaging in a wide variety of criminal activities. All cases in the OCM result from
long-term police investigations, often spanning a period of several years. The police
files constitute a rich source of data and include, amongst others, the results of special investigation techniques (e.g., wiretapping and observation) and transcripts of
suspect interrogations. These files are systematically summarised using an extensive checklist that has expanded throughout the years (Kruisbergen et al. 2012).
From a total of 180 cases in the OCM, a subset of sixteen cases was selected for
the purpose of this study. Because the focus of this study is drug-related organised
crime at major transit points in the Netherlands, all drug cases were initially selected.
284
R. Madarie and E. W. Kruisbergen
Subsequently, cases in which the airport or seaport(s) played no role, or only a
minor role, were filtered out. Other criteria employed to select cases were the richness of information with regard to trafficking operations as well as the recency of
the cases. Eleven cases concern drug trafficking through Amsterdam Airport
Schiphol (passenger as well as cargo air transport), five cases concern drug trafficking through the seaport of Rotterdam and/or other seaports in the Netherlands or
neighbouring countries.
The 16 cases together include 286 offenders. The smallest case includes four
offenders, the largest case includes 49 offenders. Four cases involve less than 10
suspects, seven cases involve 10–20 suspects, and five cases involve more than 20
suspects. The majority of offenders (90%) are male. In only one case, the main suspect was female. The timeframe in which the police investigations started ranges
from 1991 up to 2012.
As for the nature of the criminal activities, in 12 cases cocaine was imported
from South America to the Netherlands. In five cases XTC (‘ecstasy’) was exported
from the Netherlands to other countries, such as England, Australia, and the United
States of America. In two cases other drugs than cocaine and XTC were transported
as well.3
To analyse the logistics of organised crime activities, information regarding the
modus operandi of criminal groups was gathered from the cases. To analyse the
occupational and social embeddedness of organised crime activities, information on
topic of jobs, the legal work place, and social relationships were gathered. All information was subsequently categorised. Examples of categories are the type of airport
personnel involved, the type of smuggling method employed, and the reasons
offenders gave to smuggle drugs. The analyses focused primarily on the airport
cases. The seaport cases were used to enrich the data. Important differences in
results between the airport and seaport are discussed separately.
The remainder of this chapter is structured as follows: First, a brief overview of
the criminal structures is provided. Furthermore, the logistics of criminal activities
are discussed. Subsequently, the occupational and social embeddedness are discussed as well as personal reasons for offenders to become and stay involved in
organised crime. Finally, a list summarising the involvement mechanisms is provided after which the chapter concludes with a general conclusion and discussion.
Please note that these numbers do not add up to 16 because in several cases the suspects transported more than one type of drugs.
3
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
285
Empirical Results
Short Description of Criminal Networks
The criminal structures encountered in the studied cases are best described as criminal networks consisting of smaller criminal groups. Separate groups often had their
own trade, but co-operated with other groups if it helped them in their trade. The
following three cases exemplify differences in network structures.
• Case 1: several groups worked together, sometimes on ad-hoc basis, sometimes
more structurally. Each group had a core with co-offenders surrounding the core
members. Most group members were related through family ties.
• Case 2: two separate groups could be distinguished. Each group had a leader.
Only the leaders and other high ranking members in each group communicated
with each other. The two leaders, a man and a woman, once had a romance. The
romance began after the woman moved in with the man. After she became
involved in his drug trade, she involved her own family members as well, which
is how the second related group came to exist.
• Case 7: two criminal groups were initially one criminal group as nearly all members in the network were employed in the same team of cargo handlers. After a
fight between two important members, these members refused to work together
any longer and split up the network in two distinct groups. A few members of the
groups remained in touch, however, to communicate about criminal activities.
In most cases, members had several different types of tasks. For example, a
member could have executive tasks and coordinating tasks or more managerial
tasks. Most groups also had a leader or multiple leaders. Suspects knew their place
in the group. In groups that were largely based on family ties, the oldest family
member in the group was usually in charge.
Logistical Bottlenecks: Getting the Drugs Through
In this section, the logistics of drug trafficking operations at the airport are described.
Central to this section is the question how drugs can be smuggled across country
borders without the smugglers getting caught through security checks. Three types
of tactics to traffic drugs without getting caught are discerned: (1) defying security
checks, (2) avoiding security checks, and (3) neutralising security checks. When
criminal groups attempt to defy security checks, they do pass security checks but
hope not to get caught by using a method of concealment (swallowing drug capsules, for example). This can be done, in principle, without the help of corrupt airport personnel of private companies or law enforcement officers. When avoiding
security checks, smugglers do not pass security checks at all, but rather circumvent
them. The avoiding tactic is facilitated by airport personnel of private companies,
286
R. Madarie and E. W. Kruisbergen
such as cargo handlers, who abuse their privileges to ensure that a package containing drugs can leave the airport without having to pass normal security procedures.
While avoiding tactics aim at simply circumventing the security checks, neutralising tactics directly target law enforcement officers performing security checks.
Customs or other law enforcement officers are bribed so the package passes the
‘checks’.4 Before these tactics are discussed in more detail, a remark has to be made
with regard to the scope and limitations of our analyses. First, due to the sensitivity
of the topics discussed, examples of smuggling methods will be described with a
limited level of detail. Second, discussion of a specific smuggling method does not
imply that it is (still) an effective smuggling method. Although some offenders are
very innovative in their use of smuggling methods, some methods might be rather
dangerous (i.e., there is a high chance of detection). Furthermore, certain smuggling
methods discussed, either techniques to hide drugs or the abuse of privileges by
airport personnel, might now be rendered less effective because of new or increased
levels of security measures.
Defying Security Checks
The most common method to defy security controls was the use of drug mules.
Drug mules were used in eight cases. In four out of eight cases, drug mules swallowed drug capsules. In each of these four cases, the criminal group tried to smuggle
drugs into the Netherlands. In six out of eight cases, drug mules carried luggage
with drugs into the plane with them. In two cases, the criminal groups deployed
both mules that swallowed drug capsules and mules that carried luggage with drugs.
Sometimes criminal groups simply hoped that their mules would not be checked
and considered a caught mule as a business risk (case 6). Other criminal groups took
precautions to try to prevent the drugs from being noticed on scanner images.
Methods aimed at security scanners were employed to defy security checks in six
cases. For example, in case 10, drug mules swallowed coins in order to try to hide
the drug capsules they swallowed when going through a scanner at the airport. Air
freight in which drugs was hidden was frequently tampered with so as to make the
packages appear normal when passing security scanners. For example, in case 3,
carbon paper was used in boxes filled with drugs intended to make the boxes appear
empty when going through a scanner. To send larger batches of drugs, criminal
groups hid drugs in larger machines. To make transportation of these large machines
appear legitimate, criminal groups registered front companies, so they could send or
receive the products in name of the company.
4
In several cases different tactics are used.
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
287
Case 13: The criminal group transported drugs in large machines such as
woodworking machines. The machines were prepared in Europe and then
flown to South America to be filled with drugs. From there, they were transported back again to Europe either by boat or by plane. To make transportation of the machines appear legitimate, the group registered a front company
on paper specialised in woodworking. Although the machines became useless
after preparing them for drug transports, this loss of value was more than
compensated by the value of drugs transported.
Avoiding Security Checks
Security checks were mostly avoided by people working at the airport, such as cargo
handlers and technicians, who (ab)used their job-related personal credentials and
privileges for drug trafficking. For instance, in four cases, computer systems were
exploited to find flight details and, if a drug mule had been arrested, passenger
details. Flight details were consulted to acquire information about, amongst others,
at which time flights with drugs mules and drug packages would arrive, and in
which container the luggage was placed. Passenger lists were consulted to ensure
that a drug mule had indeed been arrested and had not just run away. When employees did not have access to a computer system, other employees with log-in credentials helped them out.
Also in four cases, the criminal group exploited vehicles of companies operating
at the airport to transport packages and luggage with drugs. With these company
vehicles, they could drive discretely from the airplane or basement to other places at
the airport where they would hand over the package to a co-offender, or even drive
off the airport.
Case 9: Suspect A was employed by company E and could therefore use company E’s vehicles. A drove with these vehicles on both landside and airside.
Suspect B worked at the airport for another company, but was not able to use
their vehicles at the airport. Therefore, B was asked to monitor the airport
area while A left the airport area with drugs hidden in a vehicle from company
E. Once A had left, B left the airport area as well. Together they drove to
another location where they transferred the drug packages from company E’s
vehicle to B’s car.
People working at the airport also abused their security badge for criminal purposes. The abuse of a personal security badge occurred in six cases. A personal
security badge is provided to employees working at the airport so they can move
from one section to another in order to carry out their duties. However, this freedom
288
R. Madarie and E. W. Kruisbergen
of movement also provided employees with many opportunities to engage in criminal activities.
Another recurring method to avoid security checks was the manipulation of luggage labels. In case 9, manipulated labels were sent to exporting countries where the
labels were attached to luggage with drugs. Upon arrival in the Netherlands, cargo
handlers could then easily pick up the luggage.
Most of the studied criminal groups searched within their social network for
persons with job-related privileges at the airport in order to avoid security checks.
However, the criminal group in the following case ensured that her members
acquired job-related privileges by establishing a company of their own at the airport.
Case 11: Cleaning company D was owned and run by suspect G and his son
H. With their company, they acquired the same privileges as other janitors
working at the airport. They abused their access to airplanes and information
systems of the airport. G was also in touch with different criminal groups that
wanted to use (criminal) services offered by G’s company. Furthermore, if
drug packages had to be stored safely at the airport, G sometimes hired other
persons working at the airport for 600 euros.
Next to abusing job-related privileges and credentials, criminal groups also
exploited the (technical) complexity of airplanes to avoid security checks. In two
cases (2, 11), drugs were smuggled across borders by hiding drug packages in the
airplane. The exporting group made sure the drugs were well hidden in the airplane
and sent instructions to the importing group about where to find the drugs. Sometimes
“line testing” was done by hiding a distinct object in the airplane that was to be
found by the importing group. Line testing had two purposes. First, by retrieving the
object, the exporting group knew that the instructions were received and correctly
interpreted by the importing group. Second, the exporting group was ensured that
the importing group could actually get the object out of the airplane without getting
caught. The necessity of line testing and the complexity of airplanes are best illustrated in case 2 where even the police could not find the packages with drugs hidden
by the exporting group:
Case 2: Janitors on Curacao hid drug packages in the airplane. The first time,
however, the importing group could not find the packages because they were
looking in the wrong place, just like the police. Subsequently, the janitors on
Curacao planned to send a teddy bear as a test to see if the importing group
could find the bear.
The criminal group in case 7 hid luggage with drugs in airfreight containers.
Because air freight containers are relatively large and therefore have a lot of storage
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
289
space, the importing group received instructions from the exporting group about
where the luggage with drugs was to be found.
Finally, criminal groups also attempted to transport drugs away from the airport
area at the moments of which they believed activity of security personnel was low.
Case 9: Suspect J told the police during an interrogation that luggage with
drugs was often transferred at specific times, because “you knew when there
were customs officers”. If luggage with drugs would arrive at an inconvenient
moment, then the luggage would be stored for a while. Storing luggage was
avoided at places with a lot of sniffer dogs.
Neutralising Security Checks
Security checks could be neutralised by corrupting customs or other law enforcement officers at Schiphol or at the airport of the source country to let drug mules
pass through security checks. Available information in several cases indicates that
corrupted officers might have been involved.5 For example, in one of the cases an
arrestee testified that customs officers received pictures of the mules beforehand so
the mules would be recognised during security checks. Having contacts within customs, however, does not guarantee a “smooth” operation, as some cases involve
disagreement on the amount of money a corrupted officer should receive.
Differences Between Airport and Seaport
The results regarding the modus operandi of criminal groups discussed so far apply
largely to both the airport and seaport. However, differences were also noted
between countermeasures deployed to defy or avoid security checks at the airport
and countermeasures deployed at the seaport. These differences largely pertain to
the difference in general use of the ports. The main function of the seaport is facilitating the flow of goods and not the flow of persons. Conversely, the airport has to
process an enormous amount of persons next to processing enormous volumes
of goods.
First, drugs were frequently transported by drug mules in airport cases. However,
in seaport cases, investigative units only uncovered drug transports in which drugs
were covered up by deck cargo or hidden in sea containers. In none of the seaport
cases persons were caught carrying drugs with them, even though in at least one
seaport case corrupted personnel sailing across the sea was discovered. Because
there is no (large) intercontinental flow of persons at seaports such as the ones in
Although some cases include information that civil servants, such as customs or police officers,
might be involved, the cases often lack sufficient information to prove such involvement.
5
290
R. Madarie and E. W. Kruisbergen
Rotterdam or Antwerp, criminal groups operating at seaports have to rely more on
the flow of goods. Therefore, secondly, criminal groups using seaports more often
used (front) companies to have a legitimate reason to hire sea containers or to legitimise container transport.
Required Skills and Assets
The results so far demonstrate that criminal groups mostly relied on either drug
mules or on job-related credentials, privileges, and knowledge to defy or avoid
security checks. For example, because airplanes are technically complex, group
members need to be familiar with the construction of airplanes to know where to
hide drug packages. They need to know what places are large enough to store the
packages and are relatively easy to open up without causing noticeable damage. If
criminal groups attempt to traffic drugs in luggage without accompanying passengers, they need members who can get hold of this luggage and transport the luggage
from airside to landside. In this case, the group needs members with privileges
which only people working at the airport have. In brief, most personal assets
required to traffic drugs at the airport are job-related assets of people working at the
airport. The role of these job-related assets, or more generally, the occupational
embeddedness of organised crime activities, is discussed in the following section.
Occupational Embeddedness
How do jobs facilitate organised crime? Kleemans and Van de Bunt (2008) argue
that certain jobs bring persons in a better position to contribute to organised crime
activities. Legitimate jobs of persons engaged in organised crime are often characterised by a relative high level of autonomy and mobility. Jobs with a high level of
autonomy are often found in, amongst others, companies aiming to deliver quick
services. Their workers are provided with a high level of independency and less
supervision, like janitors (at airports) and truck drivers. Truck driving is also a typical example of jobs with high mobility, which makes truck drivers more suitable to
transport illegal goods without raising suspicion. When individuals become involved
in organised crime, it is likely that their criminal activities relate to their legal jobs.
For example, a truck driver transporting legal products can easily (also) transport
illegal products. Besides, the truck driver could use his or her practical knowledge
and contacts from his or her legal activities as well (Van de Bunt and Kleemans
2007; Kleemans and Van de Bunt 2008). Not only do jobs provide the logistics and
social contacts to contribute to organised crime, having the same job also increases
trust amongst co-offenders since they know about each other’s reputation.
Furthermore, they can keep an eye on each other when working together (Kleemans
and Van de Bunt 2008). How job-related factors affect engagement in organised
crime is discussed in this section. The following five factors are elaborated upon: (1)
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
291
the similarity between legitimate job duties and criminal activities, (2) autonomy,
(3) mobility, (4) job-specific knowledge acquired through job-related privileges,
and (5) social capital of employees.
When analysing the jobs of organised crime offenders in the studied cases, a
distinction can be made between civilian personnel and law enforcement personnel.
Examples of civilian personnel encountered in the cases are stewards, janitors, technicians, and personnel engaged in (coordination of) cargo handling. Law enforcement personnel encompass customs and police officers. In the majority of cases,
civilian personnel was found to abuse job-related privileges and knowledge to facilitate drug trafficking. In one case, a member of a criminal group became employed
at the airport, but the group was arrested before he could abuse his position.
One of the job-related factors facilitating engagement in organised crime was the
similarity between licit job duties and criminal activities. Because cargo handlers
frequently have to move luggage from one place to another, moving luggage with
drugs did not raise suspicion. In a similar way, technicians have to inspect airplanes
from the inside. Therefore, moving ceilings, removing screws, and checking the
water tanks when looking for hidden drugs all appeared to be part of regular duties.
In general, it could be stated that employees with mainly executive duties in their
legitimate job also mainly carried out executive tasks in their role of criminal group
member. Employees with legitimate coordinating and supervising duties often had
more diverse types of tasks in the criminal group. For example, the director of a
cleaning company in case 11 supervised criminal group members, but also engaged
in logistic activities. A municipal clerk in case 14 gave the criminal group judicial
advice, which relates to his legitimate job, but also transported drugs a few times.
Finally, the similarity between criminal activities and legitimate job duties also
applied to law enforcement personnel. For example, in case 2, criminal group members said that they had corrupted a customs officer, so he would fake checking the
luggage of a mule.
A relative high level of mobility and autonomy are also advantageous for engagement in organised crime activities. Due to high mobility, stewards had plenty of
opportunities to traffic drugs to and from different countries themselves and they
could explore new routes to expand their drug trade. Because cargo handlers were
very mobile, they had plenty of opportunities to transport luggage with drugs from
one side to the other side of the airport without raising suspicion. A high level of
autonomy allowed janitors and technicians to smuggle drug packages from airplanes, cargo handlers to discretely separate luggage with drugs from licit luggage,
and directors to access computer systems at the airport. A high level of autonomy
also facilitated changes in work schedules advantageous to the criminal group. In
one of the cases, the manager of corrupted employees stated in an interview with the
police that the employees frequently changed their work schedule. Employees were
more or less free to switch schedules with other employees. In the following example, the coordinator of a team of cargo handlers made sure that the right employees
were scheduled to work at the right time so they could easily take out luggage
with drugs.
292
R. Madarie and E. W. Kruisbergen
Case 7: Suspect K was a coordinator at a luggage handling company at the
airport. He was therefore able to create or adjust work schedules so that
cargo handlers involved in drug trafficking would be present at the time luggage with drugs arrived. K, however, was not a high ranking member in the
criminal group. Therefore, K only initially designed the work schedule according to the criminal group’s wishes. Later, group members and cargo handlers
L and M created the work schedule themselves while K only had to turn a
blind eye.
This case also demonstrates how mobility, job-related privileges, and the similarity between job duties and criminal activities facilitate drug trafficking at the airport:
Case 7: Cargo handlers working in the luggage handling basement took luggage with drugs out of the container. One of the cargo handlers then carried
the luggage to landside while pretending it was regular luggage. At a public
road on landside, he handed the luggage over to another member of the criminal group. Because the road was busy, the luggage could discretely be transferred. The other member then drove the luggage out of the airport area.
Another job-related factor facilitating engagement in organised crime is job-­
specific knowledge acquired through job-related privileges. For instance, knowing
at which place and time luggage arrives and where that luggage has to be transported to, is job-specific knowledge. To acquire this information, cargo handlers had
to have access to computer systems in which information about the arrival of airplanes can be retrieved. Examples of other privileges resulting in job-specific
knowledge are using company’s vehicles in order to transport luggage or entering
airplanes in order to clean them on the inside. These privileges, or duties, result in
knowledge such as knowing how to manoeuvre airport vehicles and how airplanes
are (roughly) constructed. It appeared from the studied cases that such knowledge is
very useful to criminal groups attempting to smuggle drugs into airplanes and from
airplanes, with or without luggage.
Finally, social capital of employees working at the airport mainly encompasses
other contacts working at the airport or contacts that used to work at the airport. The
following case demonstrates the usefulness of contacts at the airport:
Case 2: Suspect N was supposed to pick up luggage with drugs at the airport.
This luggage, however, had already been taken away by the police. While N
was looking for the luggage, he ran into one of his contacts working at the
airport. This contact was able to access the computer information systems
and could see that the luggage had been taken away by the police. N informed
a higher ranking member, suspect O, about the police seizure.
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
293
To summarise, the occupational embeddedness of organised crime activities is
manifested through five job-related factors. Autonomy, mobility, and the similarity
between legitimate duties and criminal activities all mainly facilitate discrete
engagement in organised crime activities during work time. Job-specific knowledge
makes a person interesting to organised crime groups because this type of knowledge is rather scarce yet very useful to groups operating at logistical nodes. Similarly,
social capital acquired through work also makes a person interesting to organised
crime groups because people who already work at a logistical node are not always
easy to find. The usefulness of social capital in organised crime is further elaborated
in the next section.
Social Embeddedness
In this section, the relationship between group members and the role of settings in
which these relationships are formed are examined in greater detail. First, the role
of ethnicity and the types of relationships between offenders are discussed.
Subsequently, subcultures and offender convergence settings encountered in the
cases are discussed.
Organised crime groups cannot rely on the law or mediation to prevent (the escalation of) conflicts. Therefore, trust, proximity, and having a shared past and future
are important factors in finding new members (Beckert and Wehinger 2011;
Kleemans and Van De Bunt 1999; Loughran et al. 2013; Von Lampe and Johansen
2004). New offenders are thus often persons with whom group members already
have a social bond. Typically, offenders look for potential co-offenders in their circle of friends, relatives, and acquaintances. This often leads to greater ethnic homogeneity within criminal (sub)groups. However, as Kleemans and Van de Bunt (1999,
p. 25) note: “The basis of criminal associations is not formed by ethnicity, but by the
social relations that exist between various persons. This applies to both immigrant
and native offenders.” Social bonds are indeed an important factor in the search for
co-offenders (Morselli 2003; Beckert and Wehinger 2011).
A type of settings that foster strong social bonds concerns subcultures. Members
of subcultures in which strong cohesion is present also demonstrate high loyalty
towards the group (Kleemans et al. 2002). Certain subcultures increase the likelihood of members to engage in crime, if that subculture endorses deviating norms
that approve of, or even promote, crime (Kruisbergen et al. 2012). With the combination of a high degree of loyalty, strong cohesion, and deviating norms, subcultures
provide a pool of potential co-offenders to criminal groups. Newcomers to subcultures or groups endorsing criminal norms are then socialised into criminal practices
(Anand et al. 2005).
Another type of setting that facilitates the introduction of new members is the
so-called ‘offender converge setting’ (Felson 2006). An offender converge setting is
a place where co-offenders meet with each other or with potential co-offenders to
discuss criminal activities. Examples of such places are market places and cafés, but
294
R. Madarie and E. W. Kruisbergen
the legal workplace could also function as an offender converge setting (Kruisbergen
et al. 2012; Kleemans and Van de Bunt 2008).
In most of the studied cases, the ethnicity of group members varied greatly. In
only two cases, criminal groups were completely ethnically homogeneous. In cases
with mixed ethnicities, ethnicity was more homogeneous within subgroups than
between subgroups. This is exemplified by case 1:
Case 1: Group 1 consisted of two Lebanese men, group 2 and 3 each consisted
of two Surinamese men, and group 4 consisted of two Turkish men. There was
also a Colombian man who received drugs from a Surinamese and another
Colombian man. The drugs were subsequently sold to a Greek, a Moroccan,
an Algerian, and a native Dutch man.
Differences in tasks were also related to differences in geographical origin
between subgroups. In cases 2 and 3, for example, a subgroup originating from
Israel and a subgroup originating from the Netherlands could be discerned. In both
cases, the Israeli subgroup was more engaged in drug trade while the Dutch subgroup was more engaged in manufacturing drugs. In case 11, most members had a
Turkish background, but mainly non-Turkish members served as brokers. Brokers
are persons in a network who connect groups that would otherwise be unrelated
(Burt 1997). Through their ethnic background, non-Turkish members were better
able to connect to criminal groups in countries other than the Netherlands and
Turkey, which in turn expanded the drug trade of the whole criminal network.
One of the reasons ethnic homogeneity was greater within (sub)groups is that
offenders frequently introduced family members as new co-offenders. The two
cases in which the criminal group was completely ethnically homogeneous were
both family organisations, that is, nearly all members were relatives of each other.
As Kleemans and Van de Bunt (1999) noted, social relations are of greater importance than ethnicity in the recruitment process. The importance of social relations
over ethnicity is exemplified in case 8. A difference in ethnic backgrounds was
noted in this case between drug mules and money mules. Money mules were all
considered trustees by the leader of the criminal group. These money mules were
also closely related to his family and they had the same ethnic background as the
leader. The drug mules, on the contrary, were not closely related to the leader and
had all sorts of ethnic backgrounds. The drug mules merely engaged in drug trafficking to earn some extra money.
In most cases, 12 out of 16, family ties existed amongst members. In at least 6 of
the 12 cases, romantic partners or ex-partners co-operated in the criminal group.
Besides family ties, most criminal groups also included friendships. In most cases,
the relationship between recruiters and their friends and acquaintances dated years
back. In a few cases it was noted that proximity was also an important factor. New
group members either lived close to the recruiting group members, or they visited
the same (local) stores or restaurants as the recruiting group members. In case 5, one
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
295
of the suspects slowly got involved after she met one of the members who lived in
an apartment above the restaurant she was working for. In case 13, one of the suspects stated that he met many of the group members years ago in the restaurant he
owned back then. Finally, in case 15, suspect A met suspect B by accident when he
was shopping in suspect B’s food store.
Besides consulting family members or friends, criminal groups also attempted to
recruit persons with certain job-related privileges. In 12 out of 16 cases, group
members consulted either direct colleagues or ex-colleagues to search for new
group members.
Case 11: Suspect D was in charge of a cleaning company at the airport which
he used as a front company to cover his drug trafficking operations. D met one
of his fellow suspects a long time ago when they both owned a bar and did
business together. Later on, when D became a janitor at the airport himself,
he met another four of his fellow suspects. D got his five former colleagues
involved in his drugs operations when starting his cleaning company.
The introduction of new group members could also be facilitated by a permissive
subculture among several members of airport personnel. In one of the cases, a person involved in the criminal investigation noted that theft of travellers’ properties
and drugs trafficking frequently occurred because it was a very lucrative activity,
whereas the chance of getting caught was rather low (see also Schelvis and Lub
2008). Due to the nature of our data (i.e., mainly police files), we could not determine with certainty to what extent a permissive culture was indeed present or how
security measures could have affected the pervasiveness of such a subculture.
Offenders not operating in permissive subcultures sometimes met potential co-­
offenders in offender convergence settings. A typical offender convergence setting
is jail. In three cases, group members established contacts in jail, which they used
later when their criminal group needed new group members. In case 15, the market
place functioned as an offender convergence setting where new contacts were established. The advantage of the market place is that it attracts people who are in the
trade business and thus already possess skills and knowledge desired by criminal groups.
To conclude, persons were most likely to be introduced to criminal groups if they
already had a social bond with group members, whether it was through old friendships, jobs, or proximity. The most common type of social bonds encountered in the
cases concerned family ties and job-related ties.
296
R. Madarie and E. W. Kruisbergen
Becoming and Staying Involved in Organised Drug Trafficking
Occupational and social embeddedness are important factors in understanding the
context in which people become involved in organised crime. However, the fact that
a person is employed as a cargo handler does not automatically imply that this person wants to engage in drug trafficking. Similarly, a person does not join a criminal
group simply because he or she has a criminal niece. Therefore, in this section,
personal circumstances and motivations that cause people to become and stay
involved in organised crime are discussed.
Becoming Involved
From the studied cases, a plethora of personal motivations and circumstances could
be distilled that could lead to engagement in organised crime. The most prominent
personal motivations and circumstances are discussed here. First, in both the occupational and the social setting, criminal groups attempted to recruit new members
that were in a vulnerable position. For example, in case 7, cargo handlers who trafficked drugs looked for potential co-offenders by first getting to know their colleagues better and establishing a friendship with them. Group members then had
conversations with their colleagues about their home situation and tried to find out
how susceptible they were to engagement in organised crime. If they found a weak
spot, that colleague would subsequently be persuaded to join the criminal group.
Financial setbacks are one of the types of negative life events that often emerged
as a motivation to become involved in organised crime. In several cases, suspects
told the police that they had financial issues (in general) or that they had debts due
to a gambling or drugs addiction. Financial issues were one of the weak spots criminal groups tried to find in new members. In case 3, a group member persuaded one
of his family members who was in a bad financial situation to become involved in
organised crime.
Case 3: Suspect P told the police that he was approached by suspect Q to
become involved in drug trafficking. P states: “One night, we were having
dinner with my family and Q’s family. At some point, Q said he wanted to
discuss something with me. After dinner, Q asked me if I wanted to go abroad
to escort drug mules. He asked me how I was doing. I told him that I was in a
bad financial situation and that I agreed to go abroad.” P also stated that “he
felt he had no choice because he thought that the money would clear his debts.
Moreover, his engagement would be limited to escorting drug mules to other
countries.”
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
297
Vulnerability due to a bad financial situation was also prominent in immigrants
who had just arrived in the Netherlands. Living in poverty after migrating to the
Netherlands was recurrently claimed to be a motivation for especially those originating from the Dutch Antilles.
Case 10: Suspect Z migrated to the Netherlands, but did not obtain a steady
job and received no welfare payment. Therefore, he did odd jobs for suspect
A, such as doing the groceries and taking the children to school, but also
criminal activities related to drug trafficking. When he was released from
prison, he found a job. However, after 6 months, he became unemployed
again. Z’s financial situation went from bad to worse and he became involved
in A’s drug trafficking business again.
It was also noted that some suspects received social welfare payment in the
Netherlands while investing their drugs money in houses and expensive cars in the
Dutch Antilles. Indeed, several suspects seemed to become involved simply because
they wanted a more luxurious life.
Case 7: Suspect R told the police about his involvement in drug trafficking:
“They just asked me so that I could do more things with my children… You can
earn 6000 euros and then you start to consider your options seriously.”
Whereas suspect R seemed to suggest that the amount of money makes people
think seriously about involvement, other suspects suggested that they were in fact
blinded by the amount of money. As a suspect in case 6 stated: “I know the kind of
scene I entered and it is about a lot of money, so you know what those people do for
all that money.” Despite the fact that some suspects were actually in a bad financial
situation when they became involved in organised crime, they might have also used
this situation as a neutralisation technique for themselves or as an attempt to reduce
criminal liability when talking to the police.
In a few cases, involvement in organised crime resulted from romance. In the
following examples, a group member started a relationship with someone who was
in a bad financial situation after which the new partner slowly became involved in
the criminal group.
Case 10: As stated by a police officer: suspect X involved her partners in her
organised crime activities. When X and Y became involved in a romantic relationship, he started to work for her by carrying out illicit activities for her
criminal group. However, she did not always pay him because, as she told
him, she already paid his rent and gave him food.
298
R. Madarie and E. W. Kruisbergen
Case 2 demonstrates the social snowball method that often describes the introduction of new group members. The social snowball method is a repeated process
in which group members search for new members in their network and those new
members do the same when they have become involved in the group (Kleemans
2012). Sometimes, the snowball method was deployed to find people with specific
assets the criminal group needed for their illicit activities. Case 2 exemplifies the
snowball effect whereby (ex-)colleagues were used to find potential co-offenders.
Case 2: Suspects H and her brother I approached suspect J because they
heard that he had worked at the airport. J knew people who were still working
there and could smuggle luggage with drugs into the plane. J then connected
H and I to someone he knew who could arrange that those people would
smuggle the drugs into planes.
Finally, several suspects admitted that they initially engaged in drug trafficking
for thrill and excitement. In case 14, one suspect engaged in drug trafficking because
he was attracted to criminal activities and the lucrative business of the group. Other
suspects in this case stated that they liked to be around the highest ranking member.
Being in his presence enhanced their status in their environment. In another case,
case 7, one of the suspects told the police: “I found it very exciting because I knew
it was forbidden. On the other hand, I also knew that I could earn a lot of money
with it.” This suspect also said that he had a debt of several thousands of euros.
Staying Involved
Money and social ties are important factors for becoming involved in organised
crime. These factors are also important, however, in staying involved in organised
crime. In case 2 one of the highest ranking members, suspect B, played ‘divide and
rule’ with his group members. According to his group members, B paid them too
little and used them. Because they were dependent on his money, however, they
remained loyal to the group and to him. Many social ties existed among the members as well, ranging from direct family and family-in-law to friendships. Similarly,
in case 13, one of the suspects stated that he only continued his involvement because
he was not yet paid the full amount of money he had earned for his involvement: “If
I would receive all the money, I would not work with drugs anymore.”
Drug mules living in poverty are not only easy bait for criminal groups, but they
are also easy to keep involved. Drug mules might start smuggling to pay off their
debts or to escape poverty. However, they are also made dependent by initially lending them money which they later have to pay back by smuggling drugs. Another
tactic to ensure the mules do not run away is to wait until they have successfully
smuggled the drugs across borders and then pay them afterwards. Nonetheless,
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
299
criminal groups preferred to involve mules that were already connected to criminal
group members because, as explained before, social ties promote trust in each other.
Next to a need for money, greed was another motivation to stay involved. Case 5
exemplifies the attraction of luxury:
Case 5: The main reason for suspect S to traffic drugs was to sustain his luxurious lifestyle. S liked to be the centre of attention. It was one of his dreams to
become co-owner of a dancing club. S probably paid criminal money to fulfil
this dream as he became co-owner during the police investigation. Suspect T
told the police that S had a Ferrari and a BMW that suspect U took away from
him because S owed U money. U had a very profitable club and was so rich
that he actually did not need to traffic drugs for money.
Threats were another method deployed by criminal groups to force their members to stay involved. In several cases, higher ranking members threatened lower
ranking members, if the latter tried to leave the group or even if they complained.
Case 12: Suspect C tells the police that he had to come to a restaurant because
of rumours about unsatisfied members. In the restaurant, he met three higher
ranking members. According to C, tensions rose during the conversation and
they accused him of complaining. At some point, one of the three higher ranking members grabbed a knife on the table and said: “If something like this
happens again, I will cut your throat.”
Sometimes, threats directed at members who tried to leave the group were more
subtle. A suspect in case 9 said that, after he had made clear he wanted to leave the
group, thieves entered his house and took away many of his belongings. Another
suspect also wanted to leave because his involvement made him too nervous. He
told the police:
Case 9: “Several nights before the first shipment, I was unable to sleep. I
continued, however, because D said “You will continue”. He said that face-­
to-­face. I felt threatened and forced. Eventually, I told D a story, namely that
from February 2008 on, surveillance on the airport would be stricter. That
way I tried to quit. (…) They never threatened me, but I felt threatened because
of the way they acted. For example, I never told them where I lived, yet they
did show up at my house.”
300
R. Madarie and E. W. Kruisbergen
Group members were not only threatened with violence when they tried to leave
the group, but also with notifying the authorities of the illegal activities these group
members had already engaged in.
Case 16: Suspect B was head of a spy shop and approached suspects C and D
to offer them an IT-related job. Because both were interested, B challenged
them to demonstrate their IT-skills. The police presume that suspect B subsequently used either C and D’s demonstration or their financial rewards to
pressure them into developing malware that would be used to neutralise the
port’s digital security systems.
Finally, often a high degree of loyalty and strong cohesion exists between members of (criminal) subcultures (Kleemans et al. 2002), which makes it harder to
leave the criminal group.
In sum, money and social bonds are amongst the most motivating factors fostering involvement in organised crime. Criminal groups often rely on the snowball
method when searching for new members in their social network. Persons who are
in a bad financial situation are relatively easily recruited by criminal groups.
However, a strive for luxury might also be a reason to become involved and stay
involved in organised crime, because large amounts of money can be earned quickly
through criminal activities. Threats of violence, either implicit or explicit, were the
primary method deployed by several criminal groups to keep group members, who
attempted to leave the group, involved. However, sometimes threats were not needed
because group members felt obliged to stay due to being part of a subculture in
which social bonds are strong and group loyalty is high.
Conclusion and Discussion
Empirical Results
Based on in-depth analyses of cases from the Dutch Organised Crime Monitor, this
study explored how drug traffickers operate at important logistical nodes in the
Netherlands. Furthermore, the social and occupational embeddedness of organised
crime activities of these drug traffickers were analysed.
As for logistical operations, three types of tactics to traffic drugs without getting
caught were discerned. First, to defy security checks, criminal groups most often
deployed drug mules or methods that were supposed to avoid unusual shapes (of
drugs) on scanner images. Second, to avoid security checks at all, criminal groups
attempted to recruit persons of private companies with job-related personal credentials or privileges through which drugs could be smuggled without passing security
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
301
checks. Therefore, deliberate recruitment was aimed at persons employed at logistical nodes such as (former) colleagues. Finally, to directly neutralise the security
checks themselves, criminal groups corrupted law enforcement officers, i.e. customs or police officers. In exchange for their help, officers received money and
pictures of drug mules so they would know which persons not to check.
The occupational embeddedness of organised crime activities was manifested
through job-related factors. Autonomy, mobility, and the similarity between legitimate duties and criminal activities facilitated discrete engagement in organised
crime activities during work time. Mobility also facilitated exploration of new trafficking routes. Legitimate duties and criminal activities were more similar in
employees with executive functions than in employees with other types of functions, such as managing or coordinating functions. Social capital and job-specific
knowledge both acquired through work are two other factors that make port employees more interesting to organised crime groups. These types of contacts and knowledge are rather scarce yet very useful to groups operating at logistical nodes.
Persons with whom group members already had a social bond were more likely
to be introduced to the criminal group. The snowball effect came into play here.
Current members searched for potential offenders in their network and once
involved, the new group members did the same. The most common types of social
bonds encountered in the cases were family ties and job-related ties such as ties with
colleagues. Involvement of colleagues was fostered in case a permissive subculture
was present at the workplace which then functioned as an offender convergence
setting.
Finally, personal motivations and circumstances leading to involvement in organised crime were examined. Social bonds and money were both important in becoming involved and staying involved. The importance of social bonds for finding and
involving new group members was exemplified by the snowball effect. Whereas
persons in a bad financial situation were relatively easy to recruit, the promise of
luxury was also hard to resist by those who were not in a bad financial situation.
Greed appeared to be more often a reason to stay involved than a need for money.
Sometimes, however, group members were not immediately given the full amount
of criminal money they had earned so as to make them continue to work for the
group. Threats of violence were also frequently employed to make members who
wanted to leave stay in the group. In some cases, threats were not needed because
group members felt obliged to stay due to being part of a subculture in which social
bonds are strong and loyalty to the group is high.
Insight into Involvement Mechanisms for Organised Crime
Earlier research has demonstrated that factors related to engagement in crime differently affect organised crime offenders as compared to high volume crime offenders.
In general, positive life events such as having a job and being married are considered to have a desisting effect on engagement in crime. However, in the case of
302
R. Madarie and E. W. Kruisbergen
organised crime, having a job, a family, and other relationships could promote
involvement in criminal activities through, for example, providing opportunities to
find capable and trustworthy co-offenders. This chapter validates previous findings
on the job characteristics autonomy, mobility, and frequent social contact facilitating engagement in different types of organised crime (Kleemans and Van de Bunt
2008; Van Koppen 2013; Van Koppen and De Poot 2013). As argued by Von Lampe
(2012), research on organised crime often lacks an analysis of modes operandi of
specific types of organised crime. This chapter provided insight into the modus
operandi and logistics of transnational drug trafficking at logistical nodes.
Furthermore, this chapter analysed involvement mechanisms for this specific type
of organised crime, i.e. drug trafficking, and for people working in this specific
context, i.e. airports and seaports. By analysing results of research on organised
crime in a clearly defined context, this chapter adds in-depth knowledge to existing,
general insights into organised crime.
Enhancing (Air)port Security
This chapter showed how drug traffickers operate when they need to pass an important logistical node in the Netherlands, i.e. Amsterdam Airport Schiphol. Logistical
nodes and companies operating at the nodes are key in the prevention of organised
crime. However, decision makers involved in security issues face several dilemmas.
Three of these dilemmas are outlined here: the tensions between personnel’s autonomy and mobility on the one hand and security on the other, the tension between
swift handling procedures and security, and the corruption dilemma.
Autonomy and mobility is more or less necessary for the flexible fulfilment of
the job duties of several types of airport personnel, such as cargo handlers. However,
autonomy and mobility also create opportunities to engage in criminal activities.
Reducing the autonomy of airport personnel and increasing supervision on their
activities could reduce security risks, but could also lead to less flexible and more
time consuming procedures. This dilemma relates to the following, more general
dilemma between speed of handling procedures and security.
An obvious way to reduce opportunities for drug trafficking is to increase security checks. However, increased security checks often result in less swift handling
procedures for transportation of persons and cargo. Due to the economic importance
of and the competition within the transportation sector, speed of handling procedures is an essential asset for airports.
One possible ingredient of a policy aiming at higher levels of security is to hire
more security personnel. However, security personnel could be corrupted. It is a
security officer’s duty to encounter criminal activities and offenders, but this also
opens up opportunities to become involved in crime (Carter 1990). Furthermore,
criminal groups might put more effort in trying to neutralise (corrupt) security personnel when increased security checks make it more difficult to simply defy or
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
303
avoid those checks (see section “Logistical Bottlenecks: Getting the Drugs
Through”).
It is because of these dilemmas alongside the large demand for and supply of
drugs, the profitability of drug trade as well as the resourcefulness of offenders, that
fighting drug trafficking is a truly ‘wicked problem’ (Rittel and Webber 1973). This
does not imply that we should just take drug trafficking for granted. The dilemmas
involved in security policies, the extent of the drug trafficking problem, as well as
the potential threat that organised crime poses to society in general, require that
governments, law enforcement agencies and private companies are constantly
engaged in finding better solutions. Moreover, the transnational character of drug
trafficking requires equally ‘transnational’ countermeasures. The recent past has
demonstrated that international cooperation can produce effective policy measures,
as was the case with the pre-boarding checks at Hato International Airport on the
island of Curacao (Kleemans et al. 2010). Scientific research can contribute by providing an empirical base for policymakers.
Directions for Future Research
The introduction of highly automated processes has greatly changed the way things
work at the ports in the last decades. However, the effect of automation on the
modus operandi of criminal groups is less obvious from the studied cases. Although
some criminal groups tried to circumvent security systems or used automated processes to their advantage, automation played no significant role in the majority of
cases. The lack of abuse of automated systems in most cases could be due to the
timeframe in which investigation of these cases took place. The earliest investigation dates back to 1991, whereas the latest investigation only started in 2012.
Because some of the studied cases are somewhat dated, it is unsure to what extent
smuggling methods of currently operating criminal groups has changed. However,
the results also demonstrate behaviours that are time-persistent, such as recruitment
of persons in one’s own social network and abuse of job-related privileges and
knowledge. Nonetheless, future research should analyse new smuggling methods
and compare to what extent fundamental changes, if any, have taken place due to the
advancement of automated processes.
Another question raised by the data used concerns the generalizability of the
results. All studied cases revolve around large logistical nodes in the Netherlands or
neighbouring countries. It may well be that the size of a logistical node affects the
types of smuggling methods employed by criminal groups as well as the ease with
which these methods are employed. Alternatively, although the exact execution of
smuggling methods may differ between logistical nodes, the general involvement
mechanisms might not. Future research is therefore urged to compare smuggling
methods and involvement mechanisms between different types of logistical nodes.
Furthermore, future research should aim at expanding insight into involvement
304
R. Madarie and E. W. Kruisbergen
mechanisms by focusing on other types of organised crime and other occupational
contexts.
Finally, the present study is also limited by selection bias. Not all criminal activities reach the attention of law enforcement and not all can be solved. In this study,
only solved cases were examined. Even though the cases analysed do demonstrate
great diversity in modus operandi and the nature of crime groups, future research is
suggested to deploy different methods to research this topic. For example, researchers could interview organised crime offenders themselves. By interviewing offenders, it would be interesting to find out if and how occupational and social
embeddedness of organised crime activities differs between different types of
organised crime.
The Cases
Airport Cases
1. Marijuana, cocaine, and heroin traded by a network consisting of several groups
that used each other’s facilities and traded drugs with each other. Marijuana was
shipped in large amounts, whereas cocaine was transported in smaller amounts
by plane to the Netherlands. A janitor at the airport used his privileges to smuggle drugs from the airport, sometimes using vehicles provided by the cleaning
company. Besides luggage, other smuggling methods were swallowing drug
capsules and, in the case of shipped drugs, hiding drugs in machines.
2. Cocaine imported from South America by two related networks that are connected through their leaders. Cocaine was smuggled into and from planes by
personnel working at the airports. Corrupt customs officials received pictures
of drug mules beforehand, so they would recognise them at the airport.
3. Dutch suspects producing XTC (‘ecstasy’) and Israeli suspects exporting this
XTC. Drugs were hidden in toy boxes that were to be transported by plane.
Some flights were avoided due to the scanners used at some airports.
4. A family organisation importing cocaine from Curacao using drug mules and
air freight. In the latter case, cocaine was added to a soup and sent as a frozen
package.
5. Export of XTC to the United States of America and Canada by stewards who
took the pills with them in their personnel luggage. After surveillance had been
increased at the personnel entrance, the group bought plane tickets to get to
airside with the drugs and then hand over the luggage to the stewards.
6. Cocaine import from Curacao and the Dominican Republic by a network with
many ties to the exporting countries. Drugs were smuggled by mules, customs
officials were corrupted, and several air routes were avoided to evade security
checks.
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
305
7. Cocaine imported from South America by a network of colleagues working in
the (luggage handling) basement of the airport. Most suspects worked in the
same team and paid their coordinator to turn a blind eye on the drug trade.
Drugs were carried off the airport with vehicles available for their regular
duties.
8. Export of XTC to Asia and Oceania and import of cocaine from South America.
Money mules were close to the Israeli core members, whereas drug mules were
less close to the core and differed greatly in ethnicity. While the Israelis in this
network were mainly occupied with trade, the Dutch were mainly occupied
with producing drugs.
9. Cocaine imported by a network of colleagues working in the (luggage handling) basement of the airport. Luggage agents, transfer agents, and coordinators all engaged in the drug trade. Luggage labels were sent back to South
America for re-use and luggage reports were falsified. The group used vehicles
of airlines to transport drugs off the airport.
10. Cocaine imported from South America by a Dominican family organisation
with a single mother being in charge. She bought plane tickets for drug mules.
The drug mules swallowed capsules and coins or carried suitcases with metal to
avoid detection by security scanners.
11. Cocaine imported from Suriname and the Dutch Antilles by a group of colleagues who were employed by a cleaning company established to traffic drugs.
Through their duties, they had the privilege to enter airplanes and search for the
hidden drug packages, and to check computer systems for information about
specific flights. In addition, they sometimes borrowed airport vehicles from
persons employed by airlines.
Seaport Cases
12. Hashish trade by a network consisting of several groups that sometimes worked
together on ad-hoc basis and sometimes more structurally. Ships were bought
by front companies to transport hashish hidden in containers to and from the
Netherlands. Sometimes the hashish was not directly brought into the harbour
by ship, but first unloaded onto a smaller boat and then brought to land. After a
transport had succeeded and the ship was no longer necessary to the group, the
ship was sold again. Persons with jobs and/or skills related to sea transport were
recruited, such as captains and electricians.
13. Cocaine trade by a family organisation. The group bought machines in Israel
and the Netherlands, transported the machines as air freight to South America,
filled them with cocaine, and then transported them back to the Netherlands by
ship. Sometimes the Netherlands served as the destination point and sometimes
as a transit point to further distribute the drugs in Europe.
14. Four interrelated groups that imported cocaine and hashish, and exported synthetic drugs such as XTC. Cocaine and hashish were hidden in containers and
306
R. Madarie and E. W. Kruisbergen
shipped from South America to the Netherlands. XTC was hidden in air freight
to the United States of America. Some suspects had connections with shipping
and transportation companies. One suspect worked in the container business.
15. Cocaine exported from South America to several European ports. Certain ports
were avoided due to the type of security scanners in use there. Cocaine was
hidden inside the pallets and masked by different types of covers, such as
bananas. The main suspect had established useful connections during his time
in jail and through his trading company. Furthermore, XTC was exported to the
United States through air freight.
16. Cocaine imported from South America by a large network that hacked a track
and trace website to follow containers in which drugs were hidden. They also
digitally intercepted pin codes to pick up the containers before the legitimately
appointed truck drivers could do so. After unloading the drugs, they returned
the containers to the legitimate companies that ordered the containers in the
first place.
References
Anand, V., Ashforth, B. E., & Joshi, M. (2005). Business as usual: The acceptance and perpetuation of corruption in organizations. The Academy of Management Executive, 19(4), 9–23.
https://doi.org/10.5465/AME.2004.13837437.
Beckert, J., & Wehinger, F. (2011). In the shadow: Illegal markets and economic sociology. Socio-­
Economic Review, 11(1), 5–30. https://doi.org/10.1093/ser/mws020.
Blok, A. (2001). Honour and violence. Malden: Blackwell Publishers Inc..
Burt, R. S. (1992). Structural holes. Cambridge, MA: Harvard University Press.
Burt, R. S. (1997). The contingent value of social capital. Administrative Science Quarterly, 42(2),
339–365.
Carter, D. L. (1990). Drug-related corruption of police officers: A contemporary typology. Journal
of Criminal Justice, 18(2), 85–98. https://doi.org/10.1016/0047-2352(90)90028-A.
Chiu, Y. N., Leclerc, B., & Townsley, M. (2011). Crime script analysis of drug manufacturing
in clandestine laboratories: Implications for prevention. The British Journal of Criminology,
51(2), 355–374. https://doi.org/10.1093/bjc/azr005.
Clarke, R. V. (1997). Introduction. In R. V. Clarke (Ed.), Situational crime prevention: Successful
case studies (pp. 1–43). Guilderland: Harrow and Heston.
Cornish, D. B. (1994). The procedural analysis of offending and its relevance for situational prevention. Crime Prevention Studies, 3, 151–196.
Eski, Y., & Buijt, R. (2016). Dockers in drugs: Policing the illegal drug trade and port employee
corruption in the port of Rotterdam. Policing, 11, 1–16. https://doi.org/10.1093/police/paw044.
European Monitoring Centre for Drugs and Drug Addiction. (2015). Perspectives on drugs:
Synthetic drug production in Europe. Luxembourg: Publications Office of the European Union.
European Monitoring Centre for Drugs and Drug Addiction. (2016). Perspectives on drugs:
Cocaine trafficking to Europe. Luxembourg: Publications Office of the European Union.
Eventon, R., & Bewley-Taylor, D. (2016). An overview of recent changes in cocaine trafficking
routes into Europe. Luxembourg: Publications Office of the European Union.
Felson, M. (2006). The ecosystem for organized crime. Helsinki: European Institute for Crime
Prevention and Control, affiliated with the United Nations.
Traffickers in Transit: Analysing the Logistics and Involvement Mechanisms…
307
Kleemans, E. R. (2007). Organized crime, transit crime, and racketeering. In M. Tonry &
C. Bijleveld (Eds.), Crime and justice in the Netherlands. Crime and justice. A review of
research (Vol. 35, pp. 163–215). Chicago: The University of Chicago Press.
Kleemans, E. R. (2012). Organized crime and the visible hand: A theoretical critique on the economic analysis of organized crime. Criminology and Criminal Justice, 13(5), 615–629.
Kleemans, E. R. (2014). Theoretical perspectives on organized crime. In L. Paoli (Ed.), Oxford
handbook of organized crime (pp. 32–52). Oxford: Oxford University Press.
Kleemans, E. R., & Van De Bunt, H. G. (1999). The social embeddedness of organized crime.
Transnational Organized Crime, 5(1), 19–36.
Kleemans, E. R., & Van de Bunt, H. G. (2008). Organised crime, occupations and opportunity.
Global Crime, 9(3), 185–197. https://doi.org/10.1080/17440570802254254.
Kleemans, E. R., Brienen, M. E. I., & Van de Bunt, H. G. (2002). Georganiseerde criminaliteit
in Nederland. Tweede rapportage op basis van de WODC-monitor. Meppel: Boom Juridische
uitgevers.
Kleemans, E. R., & De Poot, C. J. (2008). Criminal careers in organized crime and social opportunity structure. European Journal of Criminology, 5(1), 69–98.
Kleemans, E. R., Soudijn, M. R., & Weenink, A. W. (2010). Situational crime prevention and
cross-border crime. In K. Bullock, R. V. Clarke, & N. Tilley (Eds.), Situational prevention of
organised crimes (pp. 17–34). Cullompton: Willan Publishing.
Kruisbergen, E. (2005). Voorlichting: Doen of laten? Theorie van afschrikwekkende voorlichtingscampagnes toegepast op de casus van bolletjesslikkers. Beleidswetenschap, 19(3), 38–51.
Kruisbergen, E. W., Van de Bunt, H. G., & Kleemans, E. R. (2012). Georganiseerde criminaliteit
in Nederland. Vierde rapportage op basis van de Monitor Georganiseerde Criminaliteit. The
Hague: Boom Lemma uitgevers.
Le, R., & Gilding, M. (2014). Gambling and drugs: The role of gambling among Vietnamese
women incarcerated for drug crimes in Australia. Australian and New Zealand Journal of
Criminology, 49, 1–18. https://doi.org/10.1177/0004865814554307.
Loughran, T. A., Nguyen, H., Piquero, A. R., & Fagan, J. (2013). The returns to criminal capital.
American Sociological Review, 78(6), 925–948. https://doi.org/10.1177/0003122413505588.
Morselli, C. (2003). Career opportunities and network-based privileges in the Cosa Nostra. Crime,
Law and Social Change, 39(4), 383–418. https://doi.org/10.1023/A:1024020609694.
Paoli, L. (2004). Italian organised crime: Mafia associations and criminal enterprises. Global
Crime, 6(1), 19–31. https://doi.org/10.1080/1744057042000297954.
Rittel, H. W. J., & Webber, M. M. (1973). Dilemmas in a general theory of planning. Policy
Sciences, 4, 155–169.
Schelvis, M. L. & Lub, A. (2008). De gelegenheid te baat nemen. Een verkennend onderzoek op
vijf Europese luchthavens naar kwetsbaarheden in het luchtvrachtafhandelingsproces die kunnen leiden tot criminaliteit. Schiphol: Expertisecentrum Luchthavens.
Schiphol. (2017). Amsterdam Airport Schiphol airport facts. Retrieved May 1, 2017 from: https://
www.schiphol.nl/nl/route-development/pagina/amsterdam-airport-schiphol-airport-facts
Tompson, L., & Chainey, S. (2011). Profiling illegal waste activity: Using crime scripts as a data
collection and analytical strategy. European Journal on Criminal Policy and Research, 17(3),
179–201. https://doi.org/10.1007/s10610-011-9146-y.
Van de Bunt, H. G., & Kleemans, E. R. (2007). Georganiseerde criminaliteit in Nederland.
Derde rapportage op basis van de Monitor Georganiseerde Criminaliteit. The Hague: Boom
Juridische uitgevers.
Van der Ploeg, J. (2014, April 2). Belangrijke cocaïneroutes verlegd naar haven
Rotterdam. De Volkskrant. Retrieved from: http://www.volkskrant.nl/magazine/
belangrijke-cocaineroutes-verlegd-naar-haven-rotterdam~a3626470/
Van Haaften, M. (2013). Tien jaar 100%-controles op Schiphol. Evaluatie rapport. Haarlem:
Bureau Discriminatiezaken Kennemerland.
308
R. Madarie and E. W. Kruisbergen
Van Koppen, M. V. (2013). Involvement mechanisms for organized crime. Crime, Law and Social
Change, 59(1), 1–20. https://doi.org/10.1007/s10611-012-9396-8.
Van Koppen, M. V., & De Poot, C. J. (2013). The truck driver who bought a café: Offenders on
their involvement mechanisms for organized crime. European Journal of Criminology, 10(1),
74–88. https://doi.org/10.1177/1477370812456346.
Van Koppen, M. V., De Poot, C. J., Kleemans, E. R., & Nieuwbeerta, P. (2010). Criminal trajectories in organized crime. British Journal of Criminology, 50(1), 102–123. https://doi.
org/10.1093/bjc/azp067.
Van Onna, J. H. R., Van Der Geest, V. R., Huisman, W., & Denkers, A. J. M. (2014). Criminal
trajectories in white-collar offenders. Journal of Research in Crime and Delinquency, 51, 759–
784. https://doi.org/10.1177/0022427814531489.
Von Lampe, K. (2012). Transnational organized crime challenges for future research. Crime, Law
and Social Change, 58(2), 179–194. https://doi.org/10.1007/s10611-012-9377-y.
Von Lampe, K., & Johansen, P. O. (2004). Organized crime and trust: On the conceptualization and
empirical relevance of trust in the context of criminal networks. Global Crime, 6(2), 159–184.
https://doi.org/10.1080/17440570500096734.
Delinquent Development, Employment
and Income in a Sample of Dutch
Organized Crime Offenders: Shape,
Content, and Correlates of Delinquent
Trajectories from Age 12 to 65
Victor van der Geest, M. Vere van Koppen, and Edward R. Kleemans
Introduction
This chapter describes the shape and content of criminal careers of a large sample
of Dutch organized crime offenders and relates these criminal careers to social economic factors such as employment and income. Other chapters in this volume provide systematic reviews of the literature regarding criminal careers of specific – largely
understudied – groups, such as criminal careers of organized crime offenders (e.g.
Savona et al. 2020). Reviews of the literature demonstrate that the vast majority of
studies focuses on juveniles, adolescents, and high-volume crime. Therefore, we do
not know whether or not ‘established facts’ about traditional criminal careers (e.g.
Farrington 2003; Blokland and Van der Geest 2017) also hold for criminal careers
in organized crime. Does a long and serious criminal career, for example, require an
early onset of offending, as is the case for traditional forms of crime? And is employment an important preventive factor for involvement in crime, as is demonstrated by
studies on employment and criminal careers (e.g. Van der Geest 2011)? Since organized crime is in many ways different from traditional forms of crime (see below),
it is important to focus more empirical research effort on this specific – seriously
understudied – group of offenders.
This chapter is based on information from five data sweeps of the Dutch
Organized Crime Monitor (OCM) including 1921 offenders, combined with information from the Dutch Offender Index (OBJD) on criminal careers of these offenders and information from Statistics Netherlands (CBS) on education, employment
and income of these offenders. After a brief overview of earlier research findings,
based on the Dutch Organized Crime Monitor, the methods section presents a
description of the data and the analyses (fixed effects and group-based trajectory
V. van der Geest (*) · M. V. van Koppen · E. R. Kleemans
Department of Criminal Law and Criminology, Faculty of Law, Vrije Universiteit
Amsterdam, Amsterdam, The Netherlands
e-mail: v.vander.geest@vu.nl
© The Author(s) 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1_13
309
310
V. van der Geest et al.
models). In the results section, we describe the criminal careers and employment of
those involved in organized crime at a particular moment in their lives. First, the
total sample is described in terms of offending, educational background, employment, and legal income. We do not only describe the criminal and employment
pathways, but also examine the effect of employment on crime. Second, we divide
the total sample in three different ways, allowing comparisons between different
subgroups of offenders. The first way is to compare criminal careers and employment for offenders involved in different types of organized crime activities. The
second way relates to offenders fulfilling different roles in a criminal organization,
whereas the third way distinguishes offenders based on their criminal trajectory.
The chapter ends with a conclusion and discussion.
Earlier Findings From the Dutch Organized Crime Monitor
The Dutch Organized Crime Monitor is an ongoing research project, carried out by
the Research and Documentation Centre, VU University Amsterdam, and Erasmus
University Rotterdam, focusing on the nature of organized crime in the Netherlands.
The main sources for the Dutch Organized Crime Monitor are closed police files of
criminal groups, often spanning a period of several years (for more information, see
Kleemans 2014). During the period 1996–2017, five data sweeps were carried out,
resulting in the systematic analysis of large-scale investigations of 180 criminal
groups involved in organized crime. In the Dutch Organized Crime Monitor, organized crime is mainly distinguished from terrorism, corporate crime, group crime,
and other types of crime by specific characteristics of the groups involved. Following
the definition of the Fijnaut Research group (Fijnaut et al. 1998), criminal groups
are considered to be organized crime groups when they are focused primarily on
obtaining illegal profits; systematically commit crimes with serious damage to society, and are reasonably capable of shielding their criminal activities from the authorities. Shielding illegal activities from the authorities is possible through various
strategies, for example: corruption, violence, intimidation, store fronts, coded communication, counter-surveillance, media manipulation, and the use of experts, such
as notaries public, lawyers, and accountants. As police priorities influence to a large
extent which criminal groups are investigated in the first place (and to which extent),
a random sample of organized crime cases is inconceivable. Therefore, the OCM
aims at a wide cross-section of empirical cases, using a strategic selection of cases
from a total population of all closed criminal investigations of national and regional
investigation teams (including the fiscal police). The 180 large-scale investigations
relate to eight different types of organized crime: (1) traditional drugs (cocaine,
heroin, cannabis) (2) synthetic drugs, (3) traditional and synthetic drugs, (4) human
smuggling, (5) human trafficking, (6) fraud and money laundering, (7) cybercrime,
and (8) other types of organized crime activities.
Furthermore, it important to note that many forms of organized crime in the
Netherlands can be characterized as ‘transit crime’ (Kleemans 2007): international
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
311
smuggling activities, such as drug trafficking, smuggling illegal immigrants, human
trafficking for sexual exploitation, money laundering and evasion of taxes (cigarette
smuggling, European Community Fraud, VAT fraud, for example), whereby the
Netherlands can be a production country, transit country, or country of destination.
Such activities pose different requirements for offenders than traditional high-­
volume crime (Kleemans and De Poot 2008). The first distinct feature of these
forms of organized crime is the greater importance of social relations, as international trade requires access to suppliers, clients, co-offenders, and profitable criminal opportunities. An important insight from earlier research on the OCM is that not
every offender has this access and that building up these relationships (including
trust) takes time and energy. The second distinct feature relates to the transnational
character of these activities. Given that most existing social networks are often
socially and geographically clustered, the trust problem in illicit trade (e.g. Reuter
1983) is even more complex and many offenders lack the necessary international
contacts that may connect them to these illegal opportunities. Finally, these activities are also logistically far more complex, e.g. transporting illegal goods over long
distances and crossing many borders without being noticed by the authorities.
Therefore, more co-offenders and specific expertise are often needed, including
contacts with the licit world, e.g. for transport, money transactions, and shielding
activities from the authorities. Many people lack these contacts and expertise, and
some acquire them only later on in life, e.g. though their professional activities and
contacts.
In their analysis of 979 suspects who were involved in 79 different organized
crime cases of the OCM, Kleemans and De Poot (2008) coined the theoretical concept ‘social opportunity structure’ – social ties providing access to profitable criminal opportunities – to explain involvement in organized crime. This concept explains
why some offenders ‘progress’ to certain types of organized crime (whereas others
do not) and why some people become involved in organized crime only later in life.
The study particularly highlighted the significant group of ‘late starters’, people
without any appreciable criminal history, and people in conventional jobs who
switch careers.
Furthermore, they analyzed the careers of 66 (ring)leaders and ‘nodal’ offenders
in more detail to find out how their criminal careers had developed. Some (ring)leaders had a long criminal career that could be characterized in four different ways: (1)
the versatile but regionally constrained ‘local hero’ (versatility as a result of both
opportunities and constraints of the local context); (2) the offenders who progressed
into organized crime through an increase in scale (specialization, particularly in
international drug trafficking); (3) those who progressed into organized crime
through capital accumulation (investing in drug transports (while refraining from
hands-on activities) and becoming a ‘background operator’); and (4) those with specific expertise, contacts, and network formation (people getting a central position in
criminal networks, because many offenders and criminal groups need their contacts
and expertise). Surprisingly, however, a significant share of (ring)leaders could be
typified as ‘late onset’ offenders, who had made a switch from a legal occupational
background to organized crime. Among these 32 offenders, a d­ istinction could be
312
V. van der Geest et al.
made between a group of 19 with a background in legal trade (including import and
export) and a group of 13 with other types of occupation – people from the business
sector, the construction industry, assembly, hotels and catering, financial services or
government. Three different types of ‘late onset’ were described: criminal activities
extending from legal activities (opportunities that arise during day-to-day work, particularly in fraud cases); people who obviously switch careers, from legal to illegal
commodities, motivated by the huge profits that can be obtained through the trade in
prohibited commodities, such as narcotics; and finally, people seizing upon criminal
opportunities only later on in life after specific, significant life events (including
financial setbacks and problematic debt situations).
Finally, they analyzed criminal careers of 92 ‘starters’, suspects who did not have
any other prior judicial contacts and who had not ‘progressed’ from high-volume
crime into organized crime. Five ‘involvement mechanisms’ were described in more
detail: (1) deliberate recruitment by criminal groups, (2) social ties and the social
snowball effect, (3) work ties, (4) leisure activities and sidelines, and (5) life events
(including financial setbacks).
A follow-up study by Van Koppen et al. (2010b), on a larger dataset, investigated
the phenomenon of ‘late starters’ in more detail. They elaborated upon this research
and investigated criminal trajectories of 854 suspects, based upon quantitative and
qualitative data from the Dutch Organized Crime Monitor. A semiparametric group-­
model was used to cluster 854 individuals into groups with similar developmental
trajectories. The most important finding of this study relates to the substantial group
of adult-onset offenders (40%) and a group without any previous criminal records
(19%), next to a group of early starters (11 percent) and a group of persisters (30%).
Up till then, no trajectory study had ever discovered such a vast share of adult-onset
offenders. Furthermore, the findings turn out to be quite robust across different
kinds of criminal activities (drugs, fraud, and other criminal activities) and different
roles in criminal groups (leaders, coordinators, and lower-level suspects). Adult-­
onset offenders emerge in several kinds of criminal activities and in several kinds of
roles in criminal groups.
Differences between the organized crime population and the general offender
population were investigated by Van Koppen et al. (2010a). Following up on the
finding that many organized crime offenders do not have judicial contacts before
adulthood, a surprising result of the comparison was that this also turned out to be
the case for the general offender population. However, organized crime offenders do
more often have previous judicial contacts, and those previous contacts are also far
more serious. Finally, these general findings turn out to be robust, as they also apply
when comparisons are made between subsets of the organized crime offenders and
the general offender population, for respectively drug crimes and fraud cases.
Van Koppen (2013) studied involvement mechanisms for organized crime in
more depth through an analysis of 15 crime groups of the Dutch Organized Crime
Monitor, including over 300 offenders. The study shows that the most common
involvement in organized crime is through family and friends who are already
involved. Almost all offender groups in the analysis contain multiple family ties
between offenders, ensuring a basis of trust. Next to family ties, the analyzed groups
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
313
also include individuals from outside the family who bring additional knowledge
and contacts. Two types of resources are explicitly discussed. On the one hand,
criminally experienced individuals who bring their criminal contacts and are used to
the level of excitement. On the other hand, offenders with conventional experience
who bring knowledge, skills and privileges gained in their occupational life. Finally,
the study highlights involvement related to negative life events: some individuals
get involved in organized crime after a financial setback or a relational crisis.
Although employment and work are often seen as preventive factors for involvement in traditional crime, the possible positive effect of employment and work relationships on involvement in organized crime is explicated in a study by Kleemans
and Van de Bunt (2008). The analysis is based upon data from 120 case studies of
the Dutch Organized Crime Monitor, involving 1623 suspects. The study describes
the different kinds of occupations encountered in cases of organized crime and the
main characteristics of these occupations. The study concludes that occupations,
work relations and work settings may provide the breeding ground for organized
crime activities, particularly transit crime. Three different ways in which occupations may present opportunities for committing organized crime are discussed.
First, through international contacts and travel movements. This could explain the
connection with occupations involving mobility, transport, and logistics: contacts
with other countries and other social groups provide ample opportunity to discover
and act upon certain opportunities of transit crime. Second, the individual freedom
of movement and/or discretion is important. This explains the involvement of directors of (small) businesses, independent professionals and, in some cases, individuals
with relative autonomy in larger organizations, such as companies and banks. More
trust and autonomy also provide more opportunity of abuse. Third, the social nature
of certain occupations is important. Occupations in which people often meet with
different people also present many opportunities to meet potential co-offenders,
buyers, or suppliers. Furthermore, the study describes various cases in more detail,
highlighting the embeddedness of certain organized crime activities in work relations and work settings. A good example is provided by work settings related to
ports, such as airport and harbors, which a very important for the import, transit, and
exports of illicit goods and/or people. In this volume, Madarie and Kruisbergen
provide an in-depth analysis of the importance of such work settings, based on
selected cases of the Dutch Organized Crime Monitor.
Methods
For the purpose of this study, information from the Dutch Organized Crime Monitor
(OCM) is combined with information on criminal careers from the Dutch Offender
Index (OBJD) and information on education, employment and income, property
values and debts from Statistics Netherlands (CBS).
314
V. van der Geest et al.
Data
Organized Crime Monitor
In five sweeps of the OCM, information of a wide cross-section of 180 cases concerning various forms of organized crime was collected. In the studied criminal
groups, 2305 offenders were involved. Police files and extensive case descriptions
allowed for classifying different roles that offenders fulfill and different types of
organized crime activities. For the current study, we distinguished three different
roles: (1) leaders, (2) coordinators, and (3) offenders with different roles.1 Leaders
are those offenders who fulfill an executive function in a criminal group; they give
orders to co-offenders and monitor the whole criminal process. Coordinators plan
and manage concrete criminal activities. For instance, they take care of offenders
performing concrete acts, such as transports of drugs. Other offenders are, for example, deployed in the actual transport of illegal goods or act as a facilitator and are
responsible for particular logistical processes. Furthermore, based on the primary
criterion offence, we distinguished between eight different types of organized
crime: (1) traditional drugs, (2) synthetic drugs, (3) traditional and synthetic drugs,
(4) human smuggling, (5) human trafficking, (6) fraud and money laundering, (7)
cybercrime, and (8) other types of organized crime activities. To make comparisons
less complex, we combined the first three categories into one broader category
(drugs). Also, since only 1% of the sample was involved in cybercrime-activities,
this category was merged with the other-category. In sum, all individuals are classified as having one of three roles and being involved in one of six types of organized
crime activities.
Dutch Offender Index
For information on offending over the life course, we used information (rap sheets)
from the Dutch Offender Index (OBJD). We were able to reconstruct the entire
criminal careers for 1921 individuals (out of 2305). These individuals constitute the
final sample for this study. For each individual in the sample, information on all
judicial contacts registered at the Dutch Public Prosecutor’s Office is available from
age 12 (the minimum age of legal responsibility in the Netherlands) up to 2016 or
death (if this occurred prior to 2016). This not only includes the timing and nature
of offenses, but also details of the individual and how the criminal case was adjudicated. Two measures of seriousness were used in the current study. First, the statutory maximum punishment under Dutch law was used, distinguished in three
different categories: (1) minor; for offences up to 4 years punishment threat, (2)
The role of 64 individuals (3,3% of the final sample) is unknown. These individuals are, therefore,
excluded from the analyses where individuals are distinguished based on their role in a criminal
organization.
1
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
315
moderate; for offences with a 4 to 8 years punishment threat, and (3) serious; for
offences related to more than 8 years punishment. This measure is independent of
the judicial history and circumstances of the suspect, but also does not take into
account the conditions under which the offences took place. Therefore, we used a
second measure of offence seriousness: the judicial settlement of cases. For each
case, it was indicated whether or not an unconditional prison sentence was imposed
and what the duration of this prison sentence was. Although this measure is not a
pure reflection of the seriousness of the offence (individual circumstances and prior
judicial contacts can affect the sentence), it does take into account the conditions
under which the offence took place.
Statistics Netherlands
For all 1921 individuals in the final sample, information on their highest education,
employment, income, property values, and debts was collected from Statistics
Netherlands. For individuals in the sample, their highest level of education attained
was measured in 2015. Unfortunately, information on education is missing for half
of the sample. Information on employment, income, property values and debts was
available for the entire sample and measured longitudinally from 1999 to 2016,
which means that it covers 18 years of each individual’s life. For privacy reasons,
frequencies lower than ten are not reported.
During the 18-year period, all employment transitions were given (dates of starting a new contract and dates of ending a contract). We used several measures to
summarize employment data. First, the distribution of working careers was given by
indicating the total amount of time someone was employed during the observation
period (never, up to half a year, half a year up to 3 years, 3–7 years, or more than
7 years). Second, λ describes the mean number of employed days in a year during
the observation period. For example, if individual A works fulltime (365 days) for
5 years and then becomes unemployed for 5 years, his λ for this 10-year period will
be 182,5. Third, for each individual, it was determined whether or not he or she was
ever self-employed during the observation period. Fourth, we used a longitudinal
dichotomous variable to indicate whether an individual was employed or unemployed during a year. An individual is considered employed in a particular year if he
or she was employed for at least 182,5 days during that year. This dichotomous
employment variable was used in all analyses examining the effect of employment
on crime.
Similar to employment, information on income is available longitudinally for
year 1999 up to year 2016. Several measures were used to examine legal earnings.
First, for each calendar year, total legal income is given in euros. Second, the income
percentile compared to the total Dutch population is given for each observed year.
Third, the proportion of years with a positive income within the observation period
is measured. Finally, it is indicated for each individual whether he or she ever
received social benefits during the observation period. It is important to note that by
using official register data, the analyses pertain to legal income. Although previous
316
V. van der Geest et al.
studies suggest that undeclared work is often conducted on a self-employed basis, it
is unknown to what extent self-employed activities are evidence for undeclared
income. Therefore, the analyses of (signals of) illegal income is beyond the scope of
the current chapter.
Several measures are used to indicate individuals’ properties and debts. First, it
was known for each year (1999–2016) what the value of the main property was.
Furthermore, the total property value gives the total value of the real estate possessed by an individual. On the other hand, most individuals also have debts. These
debts are broken down into mortgage debt and other debts.
Analyses
Simple comparisons between subgroups of organized crime offenders were carried
out. Within tables, differences between subgroups were analyzed by ANOVA’s (for
continuous variables) or Chi-square tests (for dichotomous variables). In case of a
significant difference between multiple subgroups on a continuous variable,
Bonferroni post hoc tests were conducted to indicate which subgroups differ
significantly.
Fixed Effects Models
To examine the causal effect of employment on crime, a methodological approach
should be used that rules out selection because obtaining a job or becoming unemployed does not occur at random. Employed individuals differ from individuals
without a job. A randomized experiment is the best approach to demonstrate a causal
effect, but is unrealistic in the current study. One of the next-best approaches to
control for a selection bias and demonstrate a causal effect is estimating a fixed
effects panel model. A fixed effect model is able to control for unobserved differences between individuals by only measuring within-individual change over time.
Because fixed effects models control for all time-stable differences, there is no need
to include time-stable control variables. For the total sample as well as for all subgroups based on role, organized crime type and offending pattern, a separate fixed
effect model was estimated, indicating the effect of employment on crime for each
subgroup.
Group-based Trajectory Models
Semi-parametric group-modeling was used to cluster groups of individuals with a
similar offending pattern over age (Nagin 2005; Nagin et al. 1995). Multiple trajectory analyses were carried out and both the Bayesian Information Criterion and
probabilities of group membership were used as a basis for selecting the optimal
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
317
model. Given the chosen model, individuals were assigned to the trajectory where
his or her posterior probability of group membership is highest. The trajectory
group an offender is assigned to, gives an estimation of the offending path he or she
has followed.
Results
Characteristics of the Offender Sample
Most individuals in the sample are male (91%) and a majority was born in the
Netherlands (61%, see Table 1). Five percent died before 2016, which is the end of
the observation period, at a mean age of 58 years.
Offending Over the Life Course
On average, organized crime offenders were 27 years old when their first judicial
contact took place (Table 2). On average they committed a total of 11 crimes during
their life course. For 9% of the offenders, the organized crime case is the only crime
that was registered. About one-third committed two to five crimes, about one quarter committed up to 10 crimes, and about one-third committed more than 10 crimes
during their lives. Most of these crimes are minor,2 and on average organized crime
offenders committed only one serious crime. Three out of four offenders spent some
time in prison during their criminal life-course. On average, they received three
separate prison sentences, and spent 2 years in prison. At the time of their organized
crime case, they were on average 38 years old. More than one-third was imposed a
prison sentence for their organized crime activities, and they spent an average of
10 months in prison. Offending within the sample started rising from age 12 on,
reached its peak at the relatively high age of 30, and slowly declined from that age
on (Fig. 1). Compared to the traditional age-crime curve (Gottfredson and Hirschi
1990), offending stayed relatively high and only started dropping to a lower level
from age 50 onwards.
Life-course Outcomes: Education, Employment and Income
Table 2 shows life-course outcomes for the organized crime offenders. More than
half of the sample with a known educational level, only got elementary or prevocational education. One out of ten received higher education. About a quarter of the
However, organized crime offenders were found to commit less minor and considerately more
serious offences than the general offender population (Van Koppen et al., 2010a).
2
318
V. van der Geest et al.
Table 1 Descriptives of the sample (N = 1921)
Male
Born in The Netherlands
Died before end observation period
Age of death
Country of birth
The Netherlands
Surinam
Turkey
Morocco
Former Netherlands Antilles
Other western countries
Other non-western countries
Unknown
Total criminal career
Age at first crime
Number of crimes
Violence
Property
Drugs
Other
One-shot offender
2–5 crimes
6–10 crimes
>10 crimes
Number of minor crimes (< 4 years punishment threat)
Number of moderate crimes (4–8 years punishment threat)
Number of serious crimes (> = 8 years punishment threat)
At least one prison sentence
Number of prison sentences
Time spent in prison (years)
Criterion case in organized crime
Age
Prison sentence
Length of prison sentence (years)
Role in criminal organization
Leader
Coordinator
Other role
Role unknown
Type of organized crimea
Traditional drugs
Synthetic drugs
Mean (SD) / N (%)
1738 (91%)
1166 (61%)
96 (5%)
58.1 (10.3)
1166 (61%)
149 (8%)
140 (7%)
44 (2%)
39 (2%)
98 (5%)
280 (15%)
5 (0%)
26.7 (10.3)
10.9 (11.3)
1.0 (1.9)
2.2 (4.4)
0.9 (1.3)
6.8 (7.0)
172 (9%)
593 (31%)
456 (24%)
692 (36%)
6.4 (6.7)
3.6 (5.6)
0.9 (1.2)
1430 (74%)
2.6 (3.7)
2.3 (3.7)
38.0 (9.8)
705 (37%)
0.8 (1.8)
217 (11%)
67 (3%)
1573 (82%)
64 (3%)
611 (32%)
156 (8%)
(continued)
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
319
Table 1 (continued)
Mean (SD) / N (%)
386 (20%)
115 (6%)
73 (4%)
389 (20%)
22 (1%)
169 (9%)
Traditional + synthetic drugs
Human smuggling
Human trafficking
Fraud and money laundering
Cybercrime
Other
All individuals are classified as being involved in one type of organized crime based on the primary offence in the criterion case
a
Table 2 Life outcomes
Mean (SD) / N (%)
Highest educational level
Elementary education
Prevocational education
Selective secondary education
Vocational training
Higher education
Unknown
Employed
Never during observation period
0–0.5 year
0.5–3.0 years
3.0–7.0 years
> 7 years
λ (mean number of employed days during the observation
period)
Ever self-employed during observation period
Mean income per observed year (in euros)
Mean income percentile per observed year
Proportion of observed years with positive income
Ever received social benefits during observation period
Main property value
Total property value
Mortgage debt
Debts (other than mortgage)
278 (15%)
205 (11%)
67 (4%)
236 (12%)
91 (5%)
1044 (54%)
467 (24%)
168 (9%)
388 (20%)
382 (20%)
516 (27%)
93.4 (105.5)
536 (28%)
18,447 (22,417)
39.3 (23.7)
0.8 (0.3)
624 (34%)
95,233 (227,938)
224,043 (2,067,946)
199,469 (265,435)
82,951 (1,340,081)
sample was never legally employed during the observation period. Half of the sample was employed up to 7 years (out of 18 years observation) and only one out of
four was employed more than 7 years. During the observation period, organized
crime offenders were employed on average 93 days a year (25% of the year). Given
that employment participation of the Dutch population fluctuated between 63.8 and
66.7% during the same period (1999–2016), employment in the organized crime
320
V. van der Geest et al.
0.5
0.4
0.3
0.2
0.1
0
12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64
Fig. 1 Offending over the life course
0.5
0.4
0.3
0.2
0.1
0
16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64
Fig. 2 Employment over the life course
sample can be considered relatively low (Statistics Netherlands 2018). Employment
participation is highest in early adulthood, but shows a steady decline from that
point on (Fig. 2). Already before the age of retirement, employment dropped to
10%. In the middle of the observation period (2008), employment participation
among 15–20 year-olds was 55%, topped at age 30–35 with 89%, was still 80%
among 50–55 year-olds and decreased to 33% among 60–65 year-olds (Statistics
Netherlands 2018). More than one out of four organized crime offenders were self-­
employed at some time between 1999 and 2016. No effect was found of employment on crime for the total sample (Table 3). The average income in the sample was
18,447 euros, which is at the 39th percentile. One out of three received a social
benefit at some point during the observation period. The main property value is relatively low compared to the total property value and the average mortgage debt.
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
321
Table 3 Effect of employment on crime
Control variables
Age
Age2
Employment
Employed (for at least a half year)
Log likelihood
Model 1
Full sample
B (SD)
Model 2
Full sample
.13∗∗∗ (.00)
−.00∗∗∗ (.00)
.06∗∗∗ (.01)
−.00∗∗∗ (.00)
−40,946.20
.02 (.03)
−16,540.65
†p < 0.10, ∗p < .05, ∗∗p < 0.01,∗∗∗p < .001
ffending and Life Outcomes Distinguished by Role in Criminal
O
Organization
Descriptives of Role in the Criminal Organization
Now that the characteristics, offending careers, and life outcomes of the total sample have been examined, we go into more detail and distinguish between several
subgroups within the sample. Within criminal groups, offenders can fulfill different
roles. One out of 10 offenders acts as a leader of a criminal group, 3% fulfills a
coordinating role, and 82% has another role within the criminal organization
(Table 1). The role of 64 individuals (3% of the sample) is unknown. These individuals are therefore excluded from the analyses where individuals are distinguished
based on their role in a criminal organization.
Link Between Employment and Crime: Leaders and Coordinators
Offenders with a coordinating role differ from leaders and others in most aspects
related to their criminal life course. Compared to leaders and those with another
role, coordinators are older at the time of their first crime, they commit less crimes
(they are more often one-shot offenders and less often chronics [i.e. committing
more than 10 crimes]). They also less often receive a prison sentence and spend less
time in prison (Table 4). In sum, leaders seem to have more severe criminal careers
in terms of frequency and seriousness than coordinators.
In line with these findings on the criminal careers, coordinators have better life
outcomes than leaders on many aspects. Coordinators are more often higher educated than leaders and others (Table 5). Only 4% of the coordinators was never
employed during the observation period, compared to 36% of the leaders and 24%
of the others. Also, 43% of the coordinators worked for 7 years or more, compared
to only 15% of the leaders and 28% of the others. On average, coordinators were
employed 147 days a year, compared to 55 days (leaders) and 97 days (others).
Coordinators also had a higher income and less often received social benefits.
322
V. van der Geest et al.
Table 4 Offending distinguished by role in criminal organization
Age at first crime∗∗∗
Age at criterion case
Total criminal career
One-shot offender∗∗
2–5 crimes∗
6–10 crimes
>10 crimes∗∗∗
Number of crimes∗∗∗
Violence∗
Property†
Drugs∗∗∗
Other∗∗∗
Number of minor crimes (< 4 years punishment
threat)∗∗∗
Number of moderate crimes (4–8 years punishment
threat)∗∗
Number of serious crimes (> = 8 years punishment
threat)∗∗∗
At least one prison sentence∗∗∗
Number of prison sentences∗∗
Time spent in prison (years)∗∗∗
Organized crime case
Prison sentence∗∗∗
Length of prison sentence (years)∗∗∗
Leader
Coordinator Other role
(N = 217) (N = 67)
(N = 1573)
Mean (SD) / N (%)
26.7A (9.2) 31.4B (12.4) 26.3A
(10.3)
38.4 (8.8) 38.4 (9.5)
37.9 (10.0)
9 (4%)
61 (28%)
49 (23%)
96 (45%)
12.0A
(11.5)
1.0AB (1.8)
2.6 (5.6)
1.2A (1.4)
7.2A (6.2)
6.7A (6.0)
12 (18%)
31 (46%)
16 (24%)
8 (12%)
5.6B (5.4)
0.4A (1.0)
1.2 (1.7)
0.5B (0.7)
3.5B (3.8)
3.2B (3.7)
139 (9%)
479 (31%)
378 (24%)
571 (36%)
11.1A
(11.5)
1.0B (2.0)
2.3 (4.3)
0.8C (1.3)
6.9A (7.1)
6.5A (6.8)
4.0A (6.6)
1.6B (2.4)
3.7A (5.6)
1.3A (1.3)
0.8B (0.8)
0.9B (1.3)
174 (80%)
3.0A (3.7)
4.4A (5.0)
37 (55%)
1.0B (1.5)
1.0B (1.8)
1178 (75%)
2.6A (3.8)
2.1B (3.5)
113 (52%)
2.2A (2.9)
22 (33%)
0.6B (1.4)
570 (36%)
0.7B (1.5)
Significant difference between offenders fulfilling different roles are indicated per variable
(†p < 0.10, ∗p < .05, ∗∗p < 0.01,∗∗∗p < .001). Within each row, different superscripts indicate a
significant difference between offenders with different roles (p < 0.05). For example, A and B differ
significantly; AB differs from neither A nor B
Neither significant differences between the subgroups were found in the number of
offenders that were self-employed, nor were any differences found in property values and debts between leaders, coordinators, and others.
Interestingly, both leaders and coordinators show a different pattern than is standard in the literature on the relationship between employment and crime: both for
leaders and coordinators, a significant positive effect of employment on crime was
found (Table 6, Model 3 and 4). Employment significantly increases offending with
31% for leaders and with 46% for coordinators. The only group that benefits from
employment in the sense that they commit less crimes, are those with other roles;
employment accounts for a crime reduction of 7% for this subgroup (Table 6,
Model 5).
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
323
Table 5 Life outcomes distinguished by role in criminal organizationa
Highest educational level∗∗∗
Elementary education
Prevocational education
Selective secondary education
Vocational training
Higher education
Unknown
Employed∗∗∗
Never during observation period
0–0.5 year
0.5–3.0 years
3.0–7.0 years
> 7 years
λ (mean number of employed days during
the observation period)∗∗∗
Ever self-employed during observation
period
Mean income per observed year (in
euros)∗∗∗
Mean income percentile per observed
year∗∗∗
Proportion of observed years with positive
income∗∗∗
Ever received social benefits during
observation period∗∗
Main property value
Total property value
Mortgage debt
Debts (other than mortgage)
Leader
(N = 217)
Coordinator
Other role
(N = 67)
(N = 1573)
Mean (SD) / N (%)
31 (14%)
16 (7%)
<10
14 (6%)
<10
143 (66%)
<10
<10
<10
13 (19%)
11 (16%)
31 (46%)
232 (15%)
179 (11%)
57 (4%)
200 (13%)
70 (4%)
835 (53%)
78 (36%)
18 (8%)
58 (26%)
30 (14%)
33 (15%)
54.6A (75.7)
3 (4%)
3 (4%
13 (19%)
19 (28%)
29 (43%)
147.2B (112.6)
274 (24%)
141 (9%)
298 (19%)
321 (20%)
439 (28%)
96.7C (107.3)
49 (23%)
18 (27%)
449 (29%)
14,098A
(19,156)
18.8A (20.9)
29,781B
(35,011)
37.7B (29.9)
18,573C
(22,149)
28.7C (25.3)
0.6A (0.4)
0.8B (0.3)
0.8B (0.3)
73 (37%)
10 (16%)
521 (35%)
90,651
(250,493)
192,581
(631,048)
200,457
(190,687)
45,128
(244,843)
143,489
(193,774)
217,794
(323,540)
254,321
(265,397)
54,061
(151,413)
94,601
(228,528)
232,745
(2,256,016)
196,769
(274,144)
91,432
(1,466,785)
Significant difference between offenders fulfilling different roles are indicated per variable
(†p < 0.10, ∗p < .05, ∗∗p < 0.01,∗∗∗p < .001). Within each row, different superscripts indicate a
significant difference between offenders with different roles (p < 0.05). For example, A and B differ
significantly; AB differs from neither A nor B
a
For privacy reasons, frequencies lower than ten are not reported
324
V. van der Geest et al.
Table 6 Effect of employment on crime differentiated by role in criminal organization
Control variables
Age
Age2
Employment
Employed (for at least a half year)
Log likelihood
Model 3
Leaders
B (SD)
Model 4
Coordinators
Model 5
Other role
−.06∗ (.03)
.00 (.00)
.21∗∗∗ (.06)
−.00∗∗∗ (.00)
.06∗∗∗ (.01)
−.00∗∗∗ (.00)
.27∗∗ (.10)
−1807.45
.38∗ (.18)
−426.48
−.07∗ (.03)
−13,832.14
†p < 0.10, ∗p < .05, ∗∗p < 0.01,∗∗∗p < .001
ffending and Life Outcomes Distinguished by Type
O
of Organized Crime
Descriptives of the Type of Organized Crime
Offenders do not only fulfill different roles within a criminal group, but are also
involved in different types of organized crime (Table 1). The majority (60%) was
involved in drug-related activities: traditional drugs (32%), synthetic drugs (8%) or
both (20%). Smaller numbers were involved in human smuggling (6%), human trafficking (4%), or cybercrime (1%). One out of five was involved in organized fraud.
Nine percent of the offenders was involved in other types of organized crime.
ink Between Employment and Crime: Different Types of Organized
L
Crime
On average, offenders in each subgroup were involved in crime for the first time
between age 24 and 30 (Table 7). However, drugs offenders (26 years) and those
involved in other activities (24 years), were younger at time of their first offence
than those involved in human trafficking (29 years), and fraud (30 years). At the
time of the organized crime cases, offenders involved in organized fraud (39 years)
were significantly older than those involved in human smuggling (32 years). During
their criminal career, offenders involved in organized drug-related (12 crimes) or
other activities (14 crimes) committed significantly more crimes than those involved
in human smuggling (6 crimes), human trafficking (8 crimes), or organized fraud (9
crimes). Also, they committed more moderate and serious crimes than those
involved in other organized crime activities. Relatedly, offenders involved in drug
activities in organized crime, spent more time in prison during their lives than all
other groups.
In terms of life outcomes, fraud offenders distinguish themselves from all other
organized crime offenders. Compared to offenders in other types of organized
crime activities (drugs, human smuggling, human trafficking, other activities),
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
325
Table 7 Offending distinguished by type of organized crime
Human
trafficking Fraud
Other
(N = 73)
(N = 389) (N = 191)
Human
smuggling
Drugs
(N = 1153) (N = 115)
Mean (SD) / N (%)
25.7AB
27.7AC (7.3)
(9.8)
38.9A
32.2B (8.9)
(10.1)
29.1C (9.8) 29.9C
(11.2)
35.0AB
38.6AC
(8.5)
(8.8)
24.3B
(11.1)
35.5BC
(8.9)
Total criminal career
One-shot offender∗∗∗
2–5 crimes∗∗∗
71 (6%)
18 (16%)
315 (27%) 48 (42%)
15 (21%)
23 (32%)
16 (8%)
50 (26%)
6–10 crimes†
>10 crimes∗∗∗
287 (25%) 32 (28%)
473 (41%) 16 (14%)
20 (27%)
15 (21%)
Number of crimes∗∗∗
6.0B (5.6)
7.9B (8.8)
Violence∗∗∗
11.9A
(11.4)
1.1A (2.0)
0.3B (0.7)
Property∗∗∗
2.2A (3.9)
0.9B (1.4)
Drugs∗∗∗
Other∗∗∗
Number of minor crimes (<
4 years punishment
threat)∗∗∗
Number of moderate crimes
(4–8 years punishment
threat)∗∗∗
Number of serious crimes
(> = 8 years punishment
threat)∗∗∗
At least one prison
sentence∗∗∗
Number of prison
sentences∗∗∗
Time spent in prison
(years)∗∗∗
Organized crime case
Prison sentence∗∗∗
Length of prison sentence
(years)∗∗∗
1.2A (1.4)
7.4A (7.3)
7.0A (7.0)
0.1B (0.4)
4.7B (4.8)
3.9B (4.6)
1.0ABC
0.6B (1.4)
(2.2)
1.4AB (2.3) 2.0AB
(3.5)
0.1BC (0.4) 0.4BC (1.0)
5.4AB (5.6) 5.7B (6.2)
4.5BC (5.4) 5.3BC (6.1)
3.8A (5.3)
1.8B (1.8)
2.5AB (4.0) 2.7B (4.4)
1.1A (1.3)
0.3B (0.5)
0.9AC (1.1) 0.5BC (1.1) 1.0A (1.5)
Age at first crime∗∗∗
Age at criterion case∗∗∗
904 (78%) 94 (82%)
52 (13%)
157
(40%)
73 (19%)
107
(28%)
8.6B (9.7)
2.9AC (3.8)
1.4B (1.2)
242
(62%)
1.8AB (2.6) 1.7B (3.1)
2.9A (4.2)
1.5B (2.1)
1.7B (2.8)
465 (40%) 62 (54%)
1.1A (2.1) 0.8AB (1.3)
54 (74%)
1.2B (2.5)
26 (36%) 91 (23%)
0.7AC (1.8) 0.2C (0.7)
44 (23%)
81 (42%)
13.8A
(15.1)
1.6C (2.5)
4.1C (8.1)
0.6C (1.2)
7.6A (7.3)
7.0AC (6.9)
5.9C (9.6)
136
(71%)
3.5C (5.1)
1.5B (2.9)
61 (32%)
0.4BC (1.0)
Significant difference between offenders fulfilling different roles are indicated per variable
(†p < 0.10, ∗p < .05, ∗∗p < 0.01,∗∗∗p < .001). Within each row, different superscripts indicate a
significant difference between offenders involved in different types of organized crime (p < 0.05).
For example, A and B differ significantly; AB differs from neither A nor B
326
V. van der Geest et al.
organized fraud offenders attained a higher educational level, more often were
employed and for longer periods of time, had higher incomes, less often received
social benefits, and had substantially higher property values (Table 8). Although
the absolute value of the debts of organized fraud offenders also is the highest out
of the five groups, the difference with debts of the others groups did not reach significance. Also, no effect was found of employment on crime for any of the subgroups (Table 9).
ffending and Life Outcomes Distinguished by Offending
O
Trajectories Over the Life-Course
escription of Developmental Trajectories and Their Link to Offending
D
Characteristics
A six-group model was selected as the optimal model to fit the criminal careers of
the 19133 organized crime offenders (Appendix A). Figure 3 provides a graphical
representation of the criminal careers of the six distinguished groups. Offending
characteristics of each of the six distinct groups are presented in Table 10. The first
three trajectory groups are named after the specific onset ages of the developmental
group. Their criminal pathways have similar shapes, but they started and peaked at
different stages in life. First, early-onset offenders are, together with high-frequency
offenders, the first to have started their criminal career at age 17. Around age 32,
they were again one of the youngest trajectory groups at the time of the criterion
case in organized crime. Nine out of ten early-onset offenders committed more than
10 crimes and 94% spent some time in prison during their careers. Mid-onset
offenders started offending a couple of years later, but they resemble early-onset
offenders in that they relatively often accumulated a large number of crimes (92%
more than 10 crimes) and also relatively often were convicted to prison at least once
(89%). Late-onset offenders committed their first crime only around age 28 and
were halfway their 40s at the time of their criterion case in organized crime. Three
out of four are chronics (> 10 crimes) and, on average, they also committed fewer
crimes than those with an early- or mid-onset start in crime. A fourth trajectory
group is labelled low-frequency offenders; they were 22 years old when they committed their first crime, and committed fewer crimes in their careers than the first
three groups. Only 13% of the low-frequency offenders committed more than 10
crimes. High-frequency offenders started as early as early-onset offenders, but all
committed more than 10 crimes, with an average of 47 crimes. They did not only
commit the most minor crimes, but also the most moderate and serious crimes of all
trajectory groups. All high-frequency offenders spent some time in prison and, on
average, they also spent the most time in prison. Most individuals (33%) were
The original sample contained 1921 offenders, but 8 offenders were excluded due to missing
values.
3
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
327
Table 8 Life outcomes distinguished by type of organized crime
Human
trafficking
(N = 73)
Fraud
(N = 389)
Other
(N = 191)
23 (20%)
12 (10%)
<10
12 (16%)
<10
<10
37 (10%)
28 (7%)
15 (4%)
30 (16%)
27 (14%)
<10
≈10
<10
56 (49%)
<10
<10
42 (58%)
44 (11%)
28 (7%)
236 (61%)
45 (24%)
≈10
66 (35%)
31 (27%)
24 (33%)
73 (20%)
36 (19%)
10 (9%)
27 (32%)
16 (14%)
31 (27%)
88.1AB
(108.9)
3 (4%)
19 (26%)
16 (22%)
11 (15%)
69.3A (89.1)
32 (8%)
77 (20%)
83 (21%)
124 (32%)
111.0B
(112.2)
32 (17%)
37 (19%)
35 (18%)
51 (27%)
93.0AB
(105.5)
26 (23%)
15 (21%)
129 (33%)
51 (27%)
14,824A
(11,854)
11,062A
(12,440)
25,680B
(32,413)
18,085A
(21,920)
24.3AB
(22.8)
0.8 (0.3)
17.9B (21.9) 33.0C (29.3)
0.7 (0.4)
0.8 (0.3)
28.8AC
(24.7)
0.8 (0.3)
55 (54%)
27 (43%)
95 (26%)
78 (42%)
37,138A
(91,536)
52,492AB
(147,573)
182,593
(81,950)
3764
(21,084)
37,591A
(79,939)
60,091AB
(115,348)
159,018
(121,406)
4552
(19,947)
172,584B
(370,213)
531,494B
(4,419,035)
241,483
(255,484)
239,102
(2,832,475)
86,174A
(177,562)
189,318AB
(925,298)
221,413
(262,968)
80,326
(641,198)
Human
Smuggling
Drugs
(N = 1153) (N = 115)
Mean (SD)/N (%)
Highest educational level∗∗
Elementary education
176 (15%)
Prevocational education 132 (11%)
Selective secondary
34 (3%)
education
Vocational training
128 (11%)
Higher education
39 (3%)
Unknown
644 (56%)
Employed†
Never during
303 (26%)
observation period
0–0.5 year
91 (8%)
0.5–3.0 years
228 (20%)
3.0–7.0 years
232 (20%)
> 7 years
299 (26%)
89.6A
λ (mean number of
employed days during the (103.3)
observation period)∗∗
315 (27%)
Ever self-employed
during observation
period†
16,819A
Mean income per
observed year (in
(18,737)
euros)∗∗∗
Mean income percentile
27.1A (24.0)
per observed year∗∗∗
0.7 (0.3)
Proportion of observed
years with positive
income†
369 (34%)
Ever received social
benefits during
observation period∗∗∗
Main property value∗∗∗
78,928A
(174,719)
Total property value∗
148,028A
(572,859)
Mortgage debt
179,211
(278,524)
Debts (other than
42,278
mortgage)
(443,414)
Significant difference between offenders involved in different types of organized crime are indicated per variable (†p < 0.10, ∗p < .05, ∗∗p < 0.01,∗∗∗p < .001). Within each row, different superscripts indicate a significant difference between offenders involved in different types of organized
crime (p < 0.05). For example, A and B differ significantly; AB differs from neither A nor B
328
V. van der Geest et al.
Table 9 Effect of employment on crime differentiated by type of organized crime
Model 6
Drugs
B (SD)
Control variables
Age
.04∗∗ (.01)
Age2
−.00∗∗∗ (.00)
Employment
−0.00 (.04)
Employed
(for at least
a half year)
Log
−10,231.07
likelihood
Model 7
Human
smuggling
Model 8
Human
trafficking
Model 9
Fraud
Model 10
Other offence
.03 (.05)
−.00∗∗ (.00)
.04 (.05)
−.00∗∗ (.00)
.12∗∗∗ (.02)
−.00∗∗∗ (.00)
.08∗∗∗ (.02)
−.00∗∗∗ (.00)
−.20 (.18)
.02 (.18)
−.02 (.07)
−.10 (.09)
−687.37
−512.77
−3027.56
−2050.00
†p < 0.10, ∗p < .05, ∗∗p < 0.01,∗∗∗p < .001
2
High-frequency offenders (3.9%)
Nr. of convictions
1.5
Late-onset offenders (12%)
Mid-onset offenders (9.7%)
1
Early-onset offenders (11.2%)
0.5
Low-frequency offenders (29.8%)
Sporadic offenders (33%)
0
12
20
28
36
Age
44
52
60
Fig. 3 Developmental trajectories of offending from age 12 to 65
assigned to a trajectory group named sporadic offenders. At age 37, individuals
assigned to this group had the latest onset age of all groups. One out of four sporadic
offenders had no other judicial contacts than the criterion case in organized crime
and only 1% committed more than 10 crimes in total. As we would expect from
their low-frequency offending trajectory, sporadic offenders committed significantly less crimes (3 in total) and less often spent time in prison (56%) compared to
all other offender groups.
Offenders assigned to trajectories with a lower offence frequency, seem to do
better on employment outcomes. Low-frequency offenders and sporadic offenders,
for example, are most often employed 7 years or more during the observation period,
on average have higher numbers of employed days per year during the observation
period, and also have higher incomes than individuals in most of the other trajectory
groups (Table 11). High-frequency offenders have the lowest employment outcomes, together with the early- and mid-onset offenders; they are less often
employed, have lower incomes, and more often received social benefits. No effect
for employment on crime was found for any of the trajectory groups (Table 12).
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
329
Table 10 Offending distinguished by trajectory group
Early-­
Mid-onset
onset
offenders offenders
(N = 215) (N = 186)
Mean (SD) / N (%)
16.7A (2.7) 19.6B (3.6)
Age at first
crime∗∗∗
Age at criterion
32.3A (6.7)
case∗∗∗
Total criminal career
One-shot
0 (0%)
offender∗∗∗
2–5 crimes∗∗∗
1 (1%)
6–10 crimes∗∗∗
19 (9%)
>10 crimes∗∗∗
195 (91%)
Number of
19.3A (6.9)
crimes∗∗∗
Violence∗∗∗
2.2A (2.1)
Property∗∗∗
4.6A (4.1)
Drugs∗∗∗
Other∗∗∗
Low-­
Late-onset frequency
offenders offenders
(N = 232) (N = 573)
High-­
frequency Sporadic
offenders offenders
(N = 75) (N = 640)
28.1C (6.3) 22.3D (4.7)
16.4A
(3.6)
36.8B
(8.0)
42.5C
(10.7)
37.3B (7.7) 44.7C (8.1) 32.7A (6.9)
0 (0%)
0 (0%)
7 (1%)
0 (0%)
165 (26%)
1 (1%)
14 (8%)
171 (92%)
21.6B (7.8)
3 (1%)
57 (25%)
172 (74%)
14.3C (5.9)
233 (41%)
258 (45%)
75 (13%)
6.7D (3.2)
355 (56%)
108 (17%)
4 (1%)
3.2B (2.3)
2.1A (2.2)
3.6B (3.4)
1.1B (1.5)
1.9C (2.2)
0.5C (0.8)
1.0D (1.2)
0 (0%)
0 (0%)
75 (100%)
47.3E
(16.7)
5.4C (4.6)
16.2E
(11.3)
1.8B (1.9)
23.9D
(12.3)
21.9D
(11.8)
1.2A (1.4) 1.5AB (2.0) 1.3A (1.4) 0.7C (0.9)
11.2A (5.4) 14.4B (6.0) 10.0A (5.1) 4.6C (2.7)
Number of minor 10.4A (5.5)
crimes (< 4 years
punishment
threat)∗∗∗
7.3A (5.0)
Number of
moderate crimes
(4–8 years
punishment
threat)∗∗∗
Number of serious 1.6A (1.6)
crimes (> = 8 years
punishment
threat)∗∗∗
At least one prison 203 (94%)
sentence∗∗∗
Number of prison 4.7A (3.1)
sentences∗∗∗
Time spent in
3.6A (4.2)
prison (years)∗∗∗
Organized crime case
Prison sentence
77 (36%)
Length of prison
1.0A (1.8)
sentence (years)∗
36.8E (9.9)
0.2D (0.5)
0.6D (0.8)
0.4D (0.7)
2.0E (2.0)
13.7B (5.8) 9.7A (5.0)
4.2C (2.7)
1.8E (1.9)
6.4A (4.6)
3.7B (2.9)
1.8C (1.6)
22.4D
(11.6)
1.0E (1.0)
1.5A (1.5)
1.0B (1.1)
0.6C (0.8)
3.1D (2.4)
0.5E (0.6)
166 (89%) 200 (86%) 425 (74%)
75 (100%) 361 (56%)
4.6A (3.8)
2.8B (2.6)
1.4C (1.4)
3.9A (5.2)
2.9A (4.1)
1.8B (2.8)
14.5D
(7.2)
6.8C (5.8)
1.1D (2.4)
68 (37%)
0.9A (1.9)
89 (38%)
1.0A (1.9)
224 (39%)
0.9A (1.7)
28 (37%)
1.1A (2.0)
219 (34%)
0.7A (1.6)
0.8E (0.9)
Significant difference between the trajectory groups are indicated per variable (†p<0.10, ∗p<.05,
∗∗p<0.01,∗∗∗p< .001). Within each row, different superscripts indicate a significant difference
between trajectory groups (p < 0.05). For example, A and B differ significantly; AB differs from
neither A nor B. Link between offending trajectory and employment
330
V. van der Geest et al.
Table 11 Life outcomes distinguished by trajectory group
EarlyMid-onset
onset
offenders offenders
(N = 215) (N = 186)
Mean (SD) / N (%)
Highest educational level∗∗∗
Elementary
40 (19%) 39 (21%)
education
Prevocational
22 (10%) 25 (13%)
education
< 10
< 10
Selective
secondary
education
Vocational
30 (14%) 17 (9%)
training
Higher education 12 (6%)
11 (6%)
Unknown
111 (52%) 94 (51%)
Employed∗∗∗
Never during
42 (20%) 48 (26%)
observation period
0–0.5 year
26 (12%) 24 (13%)
0.5–3.0 years
53 (25%) 59 (32%)
3.0–7.0 years
50 (23%) 40 (22%)
> 7 years
44 (21%) 25 (13%)
77.0A
λ (mean number
60.5AB
of employed days (84.5)
(77.1)
during the
observation
period)∗∗∗
55 (26%) 62 (33%)
Ever self-­
employed during
observation
period†
Mean income per 11,349
13,503
observed year (in (13,048)A (17,946)AB
euros)∗∗∗
31.3
Mean income
33.6
(19.8)A
percentile per
(21.2)AB
observed year∗∗∗
0.6 (0.3)A 0.7 (0.3)AB
Proportion of
observed years
with positive
income∗∗∗
83 (40%) 75 (41%)
Ever received
social benefits
during
observation
period∗∗∗
Late-onset
offenders
(N = 232)
Lowfrequency
offenders
(N = 573)
High-­
frequency Sporadic
offenders offenders
(N = 75) (N = 640)
29 (13%)
73 (13%)
22 (29%)
75 (12%)
19 (8%)
85 (15%)
12 (16%)
42 (7%)
< 10
< 10
< 10
< 10
20 (9%)
84 (15%)
10 (13%)
75 (12%)
15 (6%)
51 (9%)
< 10
149 (64%) 280 (49%) 30 (40%)
68 (11%)
380 (59%)
77 (33%)
106 (19%) 18 (24%)
176 (28%)
23 (10%)
50 (22%)
42 (18%)
47 (20%)
70.5AB
(95.0)
37 (6%)
108 (19%)
127 (22%)
195 (34%)
113.9C
(110.4)
22 (29%)
20 (27%)
11 (15%)
4 (5%)
35.3B
(53.5)
36 (6%)
125 (20%)
112 (18%)
201 (31%)
105.2C
(115.4)
80 (35%)
147 (26%) 17 (23%)
175 (27%)
18,722
(23,424)BC
18,271
(19,987)B
9907
(7959)A
23,576
(27,317)C
38.8
(23.5)BC
40.3
(23.7)C
27.0
(15.2)AB
44.6 (24.9)D
0.7 (0.3)BC
0.7 (0.3)C
0.7 (0.3)AC 0.8 (0.3)D
59 (26%)
196 (37%) 43 (57%)
68 (28%)
(continued)
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
331
Table 11 (continued)
Main property
value∗∗∗
Total property
value
Mortgage debt†
Debts (other than
mortgage)
EarlyMid-onset
onset
offenders offenders
(N = 215) (N = 186)
Mean (SD) / N (%)
49,961A
67,137A
(127,912) (128390)
173,493
151,485
(986,225) (571,628)
160,232
134,947
(158,204) (122,165)
75,045
29,118
(625,224) (202,548)
Late-onset
offenders
(N = 232)
Lowfrequency
offenders
(N = 573)
High-­
frequency Sporadic
offenders offenders
(N = 75) (N = 640)
110,671AB
(232,233)
248,003
(986,861)
252,657
(539,334)
108,909
(895,649)
75,344A
(170,029)
109,140
(315,236)
183,121
(169,941)
20,339
(180,173)
17,308A
(62,634)
57,805
(312,941)
92,110
(103,100)
21,023
(167,805)
141,341B
(311,585)
377,778
(3,484,204)
223,635
(231,597)
155,551
(2,235,473)
Significant difference between trajectory groups are indicated per variable (†p < 0.10, ∗p < .05,
∗∗p < 0.01,∗∗∗p < .001). Within each row, different superscripts indicate a significant difference
between trajectory groups (p < 0.05). For example, A and B differ significantly; AB differs from
neither A nor B
Table 12 Effect of employment on crime differentiated by trajectory group
Model 11
Early-­onset
offenders
B (SD)
Control variables
Age
.26∗∗∗
(.03)
Age2
−.01∗∗∗
(.00)
Employment
.02 (.06)
Employed
(for at least a
half year)
Log
−3713.70
likelihood
Model 14
Low-­
frequency
offenders
Model 15
High-­
frequency
offenders
Model 16
Sporadic
offenders
Model 12
Mid-­onset
offenders
Model 13
Late-­onset
offenders
.05∗ (.02)
.21∗∗∗ (.03) .07∗∗ (.02)
−.00∗∗∗
(.00)
.19∗∗∗
(.02)
−.00∗∗∗
(.00)
−.00∗∗∗
(.00)
.13∗∗∗
(.03)
−.00∗∗∗ (.00) −.00∗∗∗
(.00)
.02 (.07)
−.00 (.07)
−.10 (.08)
−.05 (.10)
−.01 (.09)
−2549.63
−2805.67
−2786.22
−1728.77
−2745.58
†p < 0.10, ∗p < .05, ∗∗p < 0.01,∗∗∗p < .001
Discussion
This chapter focused on the criminal careers of 1921 offenders from the Dutch
Organized Crime Monitor and analyzed delinquent development and relationships
with socio-economic variables such as employment and income. Next to many
detailed findings for various subgroups on the criterion case, development of
offending, educational level, employment and income, the most important conclusions are the following. First, organized crime offenders, on average were 27 years
332
V. van der Geest et al.
old when their first judicial contact took place and, compared to the traditional
age-crime curve (e.g. Gottfredson and Hirschi 1990), offending stayed relatively
high and only started dropping to a lower level from age 50 on. This is similar to
findings in previous studies on Dutch organized crime offenders, in which age of
onset ranges between 24–27 (Kleemans and De Poot 2008; Van Koppen et al.
2010a. Also Savona et al. (2020), find that mafia members committed their first
crime at on average age 25. Their analysis of trajectories reveals roughly similar
patterns, with a small group of persistent high-frequency offenders. In studies in
the UK and Australia, organized crime offenders are slightly younger at their first
offence, but this is also true for general offenders (Francis et al. 2013; Fuller et al.
2019). Second, no effect was found of employment on crime for the total sample.
This finding is markedly different from earlier studies on general offender samples, that identified an effect of employment on crime (Van der Geest et al. 2011;
Laub and Sampson 2003). Studies on white-collar crime offenders, however, show
that employment can also provide specific opportunities that lead offenders into
crime (Weisburd et al. 2001; Van Onna 2018). Similarly, in organized crime certain
occupations may present opportunities for committing crimes, for example through
travel movements or international contacts, and depending on the nature of the
occupation, the effect of employment on crime may be differential. Third, interesting findings emerged from analyses where individuals were distinguished based on
their role in a criminal organization. Both leaders and coordinators show a different pattern than is standard in the literature on the relationship between employment and crime: both for leaders and coordinators, a significant positive effect of
employment on crime was found. Employment significantly increases offending
with 31% for leaders and with 46% for coordinators. The only group that benefits
from employment, are those with other roles; employment accounts for a crime
reduction of 7% for this subgroup. Fourth, no effect was found of employment on
crime for subgroups distinguished by type of organized crime activity. Fifth, trajectory analyses resulted in the selection of a model with six groups: high-­frequency
offenders (3.9%), early-onset offenders (11.2%), mid-onset offenders (9.7%), lateonset offenders (12%), low-frequency offenders (29.8%), and sporadic offenders
(33%). Sixth, offenders assigned to trajectories with a lower offence frequency,
seem to do better on employment outcomes. No effect for employment on crime
was found for any of the trajectory groups.This study describing long-term criminal career characteristics and differential effects of employment on crime in a
sample of organized crime offenders, has added new findings to literature on work
and crime. Work settings may provide new opportunities for organized crime
activities, as is demonstrated by qualitative research (e.g. Kleemans and Van de
Bunt 2008; Madarie and Kruisbergen 2020). Therefore, the received wisdom that
employment has a preventive effect on crime does not seem to hold for careers in
organized crime. An interesting finding is that the effect of employment may differ
for different roles in criminal organizations. Both for leaders and coordinators, we
found that employment actually increases offending, whereas for other roles
employment led to a (small) reduction in crime. Testing to what extent the effect
also differs by the type of occupation was beyond the scope of this chapter, and
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
333
also the role of self-employment in organized crime remains largely unknown.
These findings call for more research on the link between employment and organized crime.
Acknowledgements The data used in this chapter are part of a continuing research project, the
Dutch Organized Crime Monitor, combined with information from the Dutch Judicial
Documentation System and Statistics Netherlands. We are grateful to all past and present members
of the Dutch Organized Crime Monitor research group for their efforts and particularly to Edwin
Kruisbergen for his contribution to this specific study. We also thank the members of the Recidivism
Monitor team of the Research and Documentation Centre and our contact persons at Statistics
Netherlands for their data deliveries and helpful support to this study.
Bayesian Information Criterion Values per Model
Number of groups
1
2
3
4
5
6
7
Bayesian Information
Criterion
−49,603.19
−46,259.48
−45,667.88
−45,197.78
−44,952.59
−44,834.29
−44,848.75
Zero-Inflated Poisson quadratic model with six groups
Group
Early-onset
Variable
Offenders
n
215
Estimated model parameters
Intercept
−2.19∗
Linear
2.01∗
Quadratic
−.41∗
Model characteristicsa
Mean group
.84
probabilities
Inflation
−.53∗
parameter (α)
∗
p < .05
Based on a six-group model
a
Mid-onset
Offenders
186
Late-onset
Offenders
232
Low-­
frequency
Offenders
573
High-­
frequency
Offenders
75
Sporadic
Offenders
640
−4.49∗
2.77∗
−.40∗
−7.53∗
3.36∗
−.38∗
−6.77∗
4.01∗
−.64∗
1.21∗
−.18∗
–
−8.06∗
2.96∗
−.31∗
.81
.83
.78
.91
.86
−.75∗
−.52∗
−.32∗
−.57∗
.32∗
334
V. van der Geest et al.
References
Blokland, A., & Van der Geest, V. (2017). International handbook of criminal careers and lifecourse criminology. London: Routledge.
Farrington, D. P. (2003). Developmental and life-course criminology: Key theoretical and empirical studies – The 2002 Sutherland award address. Criminology, 41, 221–255.
Fijnaut, C., Bovenkerk, F., Bruinsma, G., & Van de Bunt, H. (1998). Organised crime in the
Netherlands. The Hague: Kluwer Law International.
Francis, B., Humphreys, L., Kirby, S., & Soothill, K. (2013). Understanding criminal careers in
organised crime. Research Report, 74, Department of mathematics and statistics and department of applied social science, Lancaster University.
Fuller, G., Morgan, A., & Brown, N. (2019). Criminal histories of Australian organised crime
offenders. Trends and Issues in Crime and Criminal Justice, 567, 1–6.
Gottfredson, M. R., & Hirschi, T. (1990). A general theory of crime. Stanford: Stanford University
Press.
Kleemans, E. R. (2007). Organized crime, transit crime, and racketeering. Crime and Justice: A
Review of Research, 35, 163–215.
Kleemans, E. R. (2014). Organized crime research: Challenging assumptions and informing policy. In E. Cockbain & J. Knutsson (Eds.), Applied police research: Challenges and opportunities (pp. 57–67). London: Routledge.
Kleemans, E. R., & De Poot, C. J. (2008). Criminal careers in organized crime and social opportunity structure. European Journal of Criminology, 5(1), 69–98.
Kleemans, E. R., & Van de Bunt, H. G. (2008). Organised crime, occupations, and opportunity.
Global Crime, 9(3), 185–197.
Laub, J., & Sampson, R. (2003). Shared beginnings, divergent lives: Delinquent boys to age 70.
Cambridge, MA: Harvard University Press.
Madarie, R., & Kruisbergen, E. W. (2020). Traffickers in Transit: Analysing the logistics and
involvement mechanisms of organized crime at logistical nodes in the Netherlands: Empirical
results of the Dutch Organised Crime Monitor. In D. Weisburd, E. U. Savona, B. Hasisi, &
F. Calderoni (Eds.), Understanding Recruitment to Organized Crime and Terrorism. Cham:
Springer International Publishing.
Nagin, D. S. (2005). Group-Based Modeling of Development. Cambridge, MA: Harvard University
Press.
Nagin, D. S., Farrington, D. P., & Moffitt, T. E. (1995). Life-course trajectories of different types
of offenders. Criminology, 33, 111–139.
Reuter, P. (1983). Disorganized crime: Illegal markets and the mafia. Cambridge, MA: MIT Press.
Savona, E. U., Calderoni, F., Campedelli, G. M., Comunale, T., Ferrarini, M., & Meneghini, C.
(2020). The criminal careers of Italian mafia members. In D. Weisburd, E. U. Savona, B. Hasisi,
& F. Calderoni (Eds.), Understanding Recruitment to Organized Crime and Terrorism. Cham:
Springer International Publishing.
Statistics Netherlands. (August 13, 2018). Statline. https://opendata.cbs.nl/statline/#/CBS/nl/
Van der Geest, V. R. (2011). Working their way into adulthood: Delinquency and employment in
high-risk boys to age 32. Doctorate thesis.
Van der Geest, V. R., Bijleveld, C. C. J. H., & Blokland, A. A. J. (2011). The effects of employment
on longitudinal trajectories of offending: A follow-up in high risk youth from ages 12 to 32.
Criminology, 39(4), 1195–1234.
Van Onna, J. (2018). Blurred lines: A study of white-collar crime involvement. PhD thesis, VU
University Amsterdam: Joost van Onna.
Van Koppen, M. V. (2013). Involvement mechanisms for organized crime. Crime, Law and Social
Change, 59(1), 1–20.
Delinquent Development, Employment and Income in a Sample of Dutch Organized…
335
Van Koppen, M. V., De Poot, C. J., & Blokland, A. A. J. (2010a). Comparing criminal careers
of organized crime offenders and general offenders. European Journal of Criminology, 7(5),
356–374.
Van Koppen, M. V., De Poot, C. J., Kleemans, E. R., & Nieuwbeerta, P. (2010b). Criminal trajectories in organized crime. The British Journal of Criminology, 50(1), 102–123.
Weisburd, D., Waring, E., & Chayet, E. F. (2001). White-collar crime and criminal careers.
Cambridge, UK: Cambridge University Press.
Open Access This chapter is licensed under the terms of the Creative Commons Attribution
4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use,
sharing, adaptation, distribution and reproduction in any medium or format, as long as you
give appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter’s Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter’s Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.
Index
A
Absolute mobility, 216
Action against Terrorism Unit (ATU), 2
Activism-Radicalism-Intentions-Scales
(ARIS), 39
Addenbrooke’s Cognitive Examination-­
Revised (ACE-R), 271
Al-Qaeda and its associated movements
(AQAM), 131
Al Qaida violent extremism, 29, 30
Amsterdam Airport Schiphol, 281, 284
Anti-Mafia Investigative Directorate
(DIA), 180
Antisocial personality disorder, 269–270
Apulian mafia, 254
Article 83a, 150
Authorship, 38, 39
Autonomy, 302
B
Behavioral radicalization, 33
Bottoms-Tankebe model, 107
Brokers, 294
C
Calderoni Mean Index, 218, 227
Calderoni Rank Index, 218, 227
“Camorra”, 246, 251
Casellario dataset, 242
Cocaine, 304–306
The Code of Criminal Procedure, 150
Cognitive radicalization, 33
Conceptualization, 31
Consensus, 31
Consumption inequality, 207, 218,
227–232, 237
CONTEST, 110
Control theory, 70
Control variables, 135
Co-offenders, 12
Counter-terrorism
domestic and international
audiences, 106
dynamic relationship, 106
experimental designs, 106
investigations, 140
issues, 115
nature, 115
outcomes, 106
police legitimacy, 106
policing, 109
practices, 105, 106, 115
raid, 113
in UK, 106, 110–111
unintended negative effects and
outcomes, 105
Counter-terrorism and Security Act 2015, 110
Counter-terrorism policing, 9, 10
Crime prevalence, 124
Crime prevention, 110
Crime scripts, 279, 282, 283
Crimes of Terrorism Act in 2004, 149–151,
155, 166
Crime specialization, 124, 125
Crime-terror link, 152
Criminal activity
history, 140
prevalence, 140
Criminal behaviour, 11, 124, 140, 149, 176,
269, 273
© Springer Nature Switzerland AG 2020
D. Weisburd et al. (eds.), Understanding Recruitment to Organized Crime
and Terrorism, https://doi.org/10.1007/978-3-030-36639-1
337
338
Criminal careers, 11, 176
Italian Mafiosi, 244
Apulian mafia, 254
Camorra mafia, 251
characteristics, 254
de-escalation, 259
macro level, 241, 245, 247
mafia association, 250
meso level, 241, 246, 247
micro level, 241, 246, 247, 258
‘Ndrangheta mafia, 254
organizations, 251, 253, 254
PMM, 242, 243, 245
preliminary analysis, 258
PROTON, 241, 264, 265
recruitment, 258–261
Sicilian mafia, 251, 252, 254
variables and parameters, 253
trajectories
employment, 309, 310
OCM (see Dutch Organized Crime
Monitor (OCM))
offending trajectories (see offending
trajectories)
organized crime offenders, 309
work and crime, 332
Criminal gangs, 178
Criminal history
disadvantages, 140
non-ideological offending, 141
stigmatization, 140
Criminal involvement, 110
Criminality, 13
Criminal networks, 176, 285
Criminal offenders, 91
Criminal organizations, 179
average age, 255, 257, 264
criminal careers (see Criminal careers)
early and late recruits, 255, 257
employment (see Employment)
individuals, 332
Italian mafia organizations, 258
life outcomes, 323
mafia offenders, 254
offenders, 310
recruitment, 263
roles, 321, 322
types, Italian mafias, 250
Criminal recidivism, 89, 91, 99
Criminal record
age of onset, 125, 126
crime specialization, 124, 125
criminological investigations, 123
descriptive analysis (see Descriptive
analysis, crime rates)
Index
diversity, 124, 125
limitations, 139–140
multivariate logistical analysis, 134–136
PIRUS (see Profiles of Individual
Radicalization in the United States
(PIRUS) database)
and political violence, 123–124
pre-radicalization criminality, 126
previous criminal behavior, 123, 124
U.S. extremists (see U.S. extremists)
violent crime, 125, 126
Criminal skills, 194
Criminal trajectories, 11
Italian Mafiosi, 241, 242
early starters, 250
GBTM, 247, 248
high frequency offenders, 248
high persistence offenders, 250
juvenile delinquents, 247
low frequency offenders, 249
moderate frequency offenders, 249
population, 248
sexual crimes, 247
“Criminal training”, 259
Criminogenic factors, 49, 124
Criminological research, 196
Criminologists, 32
Criminology, 86, 121, 122, 139
Criminotrophic factors, 9, 49
D
“Dark number” issue, 197
Data sources, 40
Defying security checks, 286, 300
Delinquency, 123
Delinquent development, Dutch
organized crime
criminal organization (see Criminal
organization)
OCM (see Dutch Organized Crime
Monitor (OCM))
offending trajectories (see Offending
trajectories)
types, 324
drug-related activities, 324
employment and crime, 324, 326, 328
life outcomes, 327
offending, 325
Delinquent trajectories, see Delinquent
development, Dutch
organized crime
Democracy Index, 34
Democratic vs. non-democratic countries, 34
Dependent variables, 28, 30, 33, 34, 39, 40
Index
De-radicalization, 85, 87
Descriptive analysis, 115
crime rates
age of crime onset, 134
PIRUS, 128
pre-radicalization, 129–131
criminal activity, pre-­radicalization, 131–133
Descriptive characteristics, 61, 62
Descriptive statistics, 25, 27
Diagnostic and Statistical Manual of Mental
Disorders (DSM), 269–271
Dichotomous variables, 48
Disruptive life experiences, 167
Distal threats, 154
Domestic and international audiences, 106
Domestic terrorism cases, 87
Drug mules, 286, 298
Drug-related charges, 132
Drug trafficking
airport vs. seaport, 289, 290
Amsterdam Airport Schiphol, 284
automated processes, 303
avoiding security checks, 285,
287–289, 300
avoiding tactic, 285
crime scripts, 282, 283
criminal activities, 298, 304
criminal networks, 285
defying security checks, 286, 300
deliberate recruitment, 301
drug mules, 298
enhancing (air)port security, 302, 303
financial setbacks, 296
financial situation, 297, 301
life events, 301
logistical nodes, 282, 303
logistics, criminal activities, 284
mechanisms, 302
money, 298–300
Netherlands, 300
neutralising security checks, 289, 301
occupational embeddedness, 290–293, 301
OCM, 283
personal motivations, 296, 301
required skills and assets, 290
smuggling methods, 286, 303
social bonds, 300, 301
social embeddedness, 293–295
social snowball method, 298
social ties, 298
social welfare payment, 297
threats of violence, 301
types, 285
Drug trafficking organizations (DTOs), 10, 11,
178, 179, 181, 191, 197, 198
339
Dutch identity and society, 166
Dutch Ministry of Justice, 150
Dutch Muslim, 151
Dutch Offender Index (OBJD), 309, 314, 315
Dutch Organized Crime Monitor (OCM), 12
adult-onset offenders, 312
CBS, 315, 316
characteristics, offenders, 317
criminal careers, 311, 312
drug trafficking, 283
empirical cases, 310
employment (see Employment)
fixed effects models, 316
general offender population vs. organized
crime population, 312
group-based trajectory models, 316
individual freedom, movement and/
discretion, 313
individuals, 313
international contacts, 313
‘late onset’ offenders, 311, 312
late starters, 311, 312
life-course outcomes, 317, 319, 320
OBJD, 314, 315
OC, 310, 312
Netherlands, 310
types, 314
occupations, 313
offending, life course, 317, 320
organized crime offenders, 331
police files and extensive case, 314
police priorities, 310
social opportunity structure, 311
social relations, 311
transnational character, 311
travel movements, 313
Dutch organized crime offenders, 12
Dutch prosecution office, 155
Dutch society, 150, 167
Dynamic factor model, 207, 211, 213
E
“Early recruits”, 255–257
Economic factors, 7
Economic hardship, 167
Education, 161
Employment, 161, 162, 192
and crime, 321, 322, 324, 332
criminal organizations, 332
and income, 331
life course, 320
opportunities, 105
and work, 313
Eocio-economic factors, 10
Index
340
Escalations parameter, 245
Ethnic Minority British Election Study
(EMBES), 40
Ethnic/religious groups, 111
European context, 37, 38
European Monitoring Centre for Drugs and
Drug Addiction (EMCDDA), 280
Examined factors, 41
Exclusion criteria, 33
Experimental/quasi-experimental studies, 34, 35
Explanation-voice dimension captures, 107
Extremism, 123
attitudes and behaviour, 55
definition, 55, 71
group membership, 56
instruments, 57
non-violent, 66
and radicalization, 56
religious/ethnic, 62, 64, 68–69
right-wing, 61, 62
risk factors, 56
theories, 69
vs. society protects, 68
types, 62, 64
and violence, 67
violent, 28, 62
Extremists, language barriers in
assessments, 56
F
Financial gains, 198
Fixed effect model, 316
G
General aggression model, 70
General strain theory, 70
General theories, 70
Good Friday Agreements, 87
Grey literature, 180
Group Based Trajectory Modelling (GBTM),
247, 248
Group identity, 194
I
Identified factors, 41–43
Impartiality, 108
Inclusion criteria, 27, 28, 30–37, 46
Income inequality
and consumption inequality, 207, 218,
232, 237
SHIW, 208
vs. social mobility, 237
socio-economic mobility vs. OC, 237
Independent variables, 30
Individual-level propensity, 125
Inequality, 7
cross-sectional, 206
development, OC, 207
dynamic panel model, 217
economic conditions, 216
modified EM algorithm, 229
OC, 205, 206, 208, 218, 227
social mobility, 205
socio-economic mobility, 215, 216
Inferential statistics, 25, 27, 44
Intergroup conflict, 153
International Centre for the Prevention of
Crime (ICPC), 28
Intrusive/unpleasant practice, 116
Investigative power, 111
Islamic extremist offenders, 89
Islamic radicalization, 9
Islamic State of Iraq and Syria (ISIS), 105
Israel Prison Service (IPS)
age at time of release, 92
classification, 90
definition, 90
history of prior security convictions, 91
incarceration and individual level risk
factors, 91
ISA, 90
length of sentence served, 92
marital status, 92
National security offences, 91
offence type, 92
organizational affiliation, 92
PA, 91
security prisoners, 90
Israeli Security Agency (ISA), 90
Italian criminal organizations, 264
Italian mafia organizations, 253, 254, 258
Italian National Anti-Mafia and
Counterterrorism Directorate
(DNA), 180
Italian Parliamentary Anti-Mafia Commission
(CPA), 180
J
Jihadists, 148
Juvenile offending, 138, 139
K
Kaplan-Meyer method, 93
Index
L
“Late recruits”, 255–257
Legal authorities, 107
Legitimacy, 106, 107, 109, 111
Life-course outcomes, OC offenders, 317, 319
“Line testing”, 288
Literature
authorship, 38, 39
data sources, 40
dependent variables, 39, 40
European context, 37, 38
examined factors, 41
inclusion criteria, 35
meta-analysis, 43, 44
methodologies, 41
primary limitations, 25
proportions, 38
publications by year, 36, 37
qualities and characteristics, 26
quality and trajectory, 27
quantitative studies, 25
radicalization (see Radicalization)
Logistic regression models, 41, 158, 159
Lone-actor’ terrorism, 56
Low self-control acts, 13
Lower super output areas (LSOA), 112
Low-level criminal activity, 85
M
Mafia, 175, 178–181, 191–194, 196–198
criminal careers (see Criminal careers)
criminal organization, 254, 255, 257, 258
developmental trajectories, 250
PMM (see Proton Mafia Members
dataset (PMM))
types, 241, 243
“Mafia Association”, 263
Marital status, 99
Masochistic personality disorder, 272, 273
Mental health, 105
Mental illness, 135
Meta-analysis, 43, 44, 48
Millon Clinical Multiaxial Inventory-III
(MCMI-III), 270, 271
Mobility, 301, 302
Money-laundering, 195
Money mules, 294
Multi-dimensionality, legitimacy, 107
Multivariate logistical analysis
control variables, 135, 137
criminal history and participation, 136, 137
dependent variable, 134
descriptions, 134
341
independent variable, 134
juvenile offending, 138, 139
missing data, 136
variables, 136
Muslim extremism, 167
N
National Security, 87, 90–92, 96, 99
Netherlands, 12, 315, 316
drug trafficking (see Drug trafficking)
functions, 12
logistical nodes, 278, 280–282, 300,
302, 303
transit crime (see Transit crime)
transnational organised crime, 278
Neuropsychological screening, 271
Neuroscience, 269, 273
Neuroscientific tools, 8, 176
Neutralising security checks, 289, 301
Non-ideological crime, 122, 125, 126, 128,
130, 133
Non-ideological criminal behaviour, 123, 124
Non-ideological criminality, 122
Non-ideological delinquency, 129
Non-ideological predictors, 122, 138, 139
Non-organized crime (NOC), 269–272
Non-violent crimes, 122, 129
Non-violent extremism, 136
Non-violent extremists, 124, 128
Non-violent offences, 89
Non-violent offenders, 99
Non-violent terrorism offenders, 89
Non-violent vs. violent political extremists, 129
Non-Western immigrants, 149
O
Occupational embeddedness, 12
Offender convergence settings, 280, 293, 295
Offending trajectories
characteristics, 326
development, 328
and employment, 328, 331
life outcomes, 330–331
Operationalization, validity, 31
Opposition to the West, 30
Ordinary least square regressions, 41
Organised crime (OC)
co-offenders, 12
occupational embeddedness, 12
port employees, 12
social bonds, 12
types of tactics, 12
Index
342
Organization for Security and Co-operation in
Europe (OSCE), 2
Organizational affiliation, 90, 92
Organized crime (OC)
activities, 3
air and sea ports, 14
competitive economies, 2
concept, 198
consumption inequality, 218, 227
criminal careers, 278
criminal history, 14
definition, 177–178, 198
economic risk factors, 177
economy, 2
factors, 212–214
family ties and jobs, 277
illegal enterprises, 1
implications, 199
inequality, 218, 227
latent factor, 211
measures, 227
Calderoni Mean Index, 211
Calderoni rank Index, 211
dynamic factor model, 207, 211, 213
indirect, 212
type I and II crimes, 208
variables, 210
mechanisms, 278
multinational governmental bodies, 2
Netherlands (see Netherlands)
OCGs (see Organized crime
groups (OCGs))
organizations, 2
P90/P10 index, 218
painstaking empirical research, 3
personality (see Personality)
PROTON, 4, 13
recruitment, 4, 175–177
risk factors, 13
social embeddedness, 175
social mobility, 11
social ties, 175
socio-economic factors, 13
socio-economic mobility, 227
transit crime (see Transit crime)
transnational organised crime, 278
Organized crime groups (OCGs), 10, 11
age, 190
consolidation, 175
cooperating/co-offending, 177
criminal background and skills, 195
and criminal careers, 176
criminal co-operations, 176
data collection, 181
DTOs, 178
economic conditions, 192, 193
economic risk factors, 176
educational background, 192
emergence, 176
employment, 192
ethnicity, 191
factor categories per type of study,
189, 190
full-text screening, 179
gender, 191
and geographic area, 197
group identity, 194
identification, 189, 197
limitations, 198
mafias, 178
members, 197
methods of analysis, 181
mixed-methods studies, 188
opportunities, 176
psychological factors, 176, 194, 195
psychological traits, 176
qualitative studies, 184–187
quantitative studies, 181–183, 197
resilience, 175
silence/omertà, 196
skills/expertise, 176
social factors, 175
social ties, 193, 194
socio-economic and psychological
factors, 189
source types, 180
vs. street gangs, 178
types, 178, 197, 198
P
PACE Code of Ethics, 111
Palestinian Authority (PA), 91, 92, 100
Palestinian organizations, 90
Parties/voters, 31
Penal Code, 150
Perceived threats, 153
psychological and criminological
notions, 149
types, 153
Personality
disorders, 270
DSM-5, 270
logistic regression analysis, 272
masochistic behaviours, 272
masochistic personality disorder, 272
masochistic profile, 273
MCMI-III, 271
Index
multifactorial construct, 269
neuropsychological tests, 271
neuropsychological/cognitive profile, OC
members, 273
OC vs. NOC prisoners, 271
participants, 270
profile, OC members, 272
psychological/behavioural
characteristics, 273
relational attachment, 272
statistical analyses, 271
Personality-trait related factors, 13
Personal Records Database (BRP), 155
PEW Research Centre, 150
Police anti-terror practices, 109
Police behaviour, 109
Police legitimacy, 62, 106
Police raids, 113
Policy implications, 139
Policy makers, 140
Policy making process, 177
Political extremism, 121, 122, 130, 139–141
Political turbulences, 150
Political violence, 127
in US, 125
Port employees, 12
Port of Rotterdam, 282
Post-radicalization violent extremism,
124, 126
Predictive quality, satisfactory level, 32
Predictors, 33
Pre-post group comparison, 61
Pre-radicalization crime rates,
U.S. extremists
criminal histories, 129–131
in non-violent crimes, 129
non-violent vs. violent political
extremists, 129
in PIRUS database, 129, 130
Pre-radicalization
criminal activity, U.S. extremists
criminal behaviours, 131–133
criminal histories, 132
PIRUS data, 132, 133
violent extremists, 132
Pre-radicalization criminal behaviour, 122
Pre-radicalization criminality, 126
Previous criminal behaviour, 123, 124
Prior criminality, 14, 15
Prior security convictions, 91, 93, 94
Prison-based programs, 88
Prisoners, 270
Proactive contacts, 111
PROBIT models, 41
343
Procedural (in)justice
Bottoms-Tankebe model, 107
component, 109
contextual influences, 109
correlational studies, 108
counter-terrorism practices, 115
crime prevention, 110
criminal involvement, 110
in criminology, 107
decision-making/judgment-making
processes, 112
description, 107
evidence base, 109
explanation-voice dimension captures, 107
factorial surveys, 112
high procedural justice raid condition, 113
impartiality, 108
in-situ application, 116
legal authorities, 107
legitimacy, 107
low procedural justice condition, 113
LSOA, 112
participant, 112
perceived risk of recruitment, 113
person’s membership in society, 108
police raids, 113
post-facto strategy, 116
potential differences, 116
principles, 116
quality of decision-making, 107
quality of interpersonal treatment, 107
security screening tactics, 109
social inclusion, 109
statistics, experimental condition, 114
street interaction, 112
street stop, 113
terrorism policing, 109
treatment, 108, 109
trustworthy motives, 108
t-tests, 114, 115
various structural and individual
factors, 105
Procedural justice script, 115
Procedural justice theory, 70
Procedurally unjust street-level
encounters, 115
Profiles of Individual Radicalization in the
United States (PIRUS) database,
122, 123, 126–134, 136–139
Protective factors, 9, 15, 89
characteristics, 58
classification, 62
conceptual and methodological
discussions, 58
Index
344
Protective factors (cont.)
data base, 70
de-radicalization, 57–59
disengagement, 57–59
double face, 67
individual and social resources, 57
investigated, 59
knowledge, 57
limitations, 70, 71
vs. political and religious
extremism, 62–64
radical attitudes, 43, 44
radical behaviours, 44, 47
radical intentions/willingness, 43, 46
vs. radicalization, 62–64
and resilience, 56–58
self-control, 66, 67
societal connectedness, 48
specification, 59
PROTON, 4
Proton Mafia Members dataset
(PMM), 11, 244
Casellario dataset, 242
definition, 243
duration, 244
escalation, 245
frequency, 244
mafia association, 243
participation, 243
socio-demographic data, 243
socio-demographic variables, 263
specialization, 244
variable, 244, 245
Psychological factors, 6, 8, 10, 152
Psychology, 269, 273
Psychopathy, 176
Public and violent anti-demonstrations, 151
Public disorder offences, 91, 92
Q
Qualitative assessments, 101
Qualitative methodologies, 26
Qualitative/mixed-methods approaches, 180
Qualitative studies
coding, 60, 61
cost-benefit-analysis, 66
de-radicalization, 59, 62, 65, 66, 69
disengagement, 61, 62, 65, 66, 69
guidelines, 59
limitations, 70, 71
in Netherlands
adverse experiences, 163
codes, 161
coping strategies and narratives,
164, 165
detainees, 161
early family life, 163, 164
education, 161
employment, 161, 162
housing, 162
identity and belonging, 164
informants, 160, 161
instrument, 160
perceived threat (see Perceived threat
and individual involvement)
religion (Islam), 165
search for belonging, 164
search for meaning, 164
socio-economic and personal
adversities, 160
socio-economic variables, 162, 163
terrorism units, 160
push and pull factors, 66, 69
self-reported disengagement, 62
Quality of marriage, 89
Quantitative methods, 26, 33
Quantitative research, 26, 47
Quantitative studies, 161, 166, 168
characteristics, 60, 61, 72
coding, 60, 61
community
society factors, 68, 69
society level, 64
family factors, 67
family level, 62
identification and synthesis, 33
individual factors, 67
individual level, 62
limitations, 70, 71
origin, SES and criminality, 156
peer group factors, 68
peer group influences, 64
protective factors, 62–64
school factors, 68
school level, 62
self-control, 66
socio-economic determinants (see
Socio-economic determinants)
statistical tests, 62
R
Radical extremism, 141
Radicalization, 121, 150, 152, 153, 167
agents, 28
Al Qaida violent extremism, 29, 30
attitudes and behaviour, 55
Index
beliefs, 9
characteristics, 28
community related factors, 29
conception, 30, 31, 33
definition, 28, 31, 55
dependent variables, 30
development, 26
factors, 29
FBI definition, 121
goal, 5
independent variables, 30
individual-level factors, 29
instruments, 57
macro and exo-systemic factors, 29
phases, 56, 66, 69
propensity, 71
protective factors, 9
psychological factors, 6
qualitative methodologies, 26
quantitative research, 26
recruitment, 5
relational/group based factors, 29
research, 31–32
risk factors, 26, 46
vs. shields of resilience, 70
social prevention, 28
socio-economic status, 6
subjective beliefs, 6
sub-literature, 26
support for terrorism, 30
systematic reviews, 26, 27
terrorism literature, 27, 28
type, 64
and violent extremism (see Violent
extremism)
in western countries, 33
Recidivism, 9
advantages, 90
affiliation, 97
age at first incarceration, 94, 95
age at time of release, 94, 95
analysis, 93
characteristics, 90
criminology, 86
declining rates, 98
incapacitation period, 98
incarceration, 98
IPS (see Israel Prison Service (IPS))
length of sentence served, 95, 96
marital status, 97
marker, 85
measure, 86
observation period, 87
offence type, 96
345
offenders, 85
organizational affiliation, 100
prior security convictions, 93, 94
pull and push factors, 88
qualitative assessments, 101
rates, 87, 88, 93
recidivism rates, 86
and recidivism-related risk factors, 97
risk factors, 85, 86, 88–90, 99, 101
scholars, 88
security offenders, 98
statistics, 85, 87
type of offending, 86
Recruitment
career criminals, 7
co-offenders, 7
criminal expertise, 14
criminal records, 14
economic factors, 7
individual-level risk factors, 4
inequality, 7
involvement mechanisms, 7
low self-control acts, 13
neuroscientific tools, 8
OC, 3, 175–177
OCGs, 10, 11
police agencies, 4
protective factors, 9
psychological factors, 8
and radicalization, 5, 6
risk factors, 15
self-reinforcing mechanisms, 7
situational factors, 4
social factors, 6, 7
social mobility, 7
socio-demographic factors, 4
terrorism, 4–6
work relations, 7
work settings, 7
Regression-based multiple imputation
operates, 136
Regression models, 41, 46
Relative mobility, 216
Religious affiliation, 149
Resilience
healthy development, 57
and protective factors, 56–58
shields of resilience, 66, 69, 70
vs. violent, 62
Right-wing authoritarian personality
(RWA), 31
Right-wing extremism, 62
Right-wing extremists, 151
Risk assessment, 57, 101
346
Risk factors, 176, 177, 180, 196, 199
background characteristics and
demographic factors, 48
combined and cumulative effect, 28
framework, 32
individual-level, 35
opposite/reverse, 48
quantitative, 57
radical attitudes, 45
radical behaviors, 33, 44, 47
radical intentions/willingness, 43, 46
radicalization, 26
radicalizing agents, 28
recidivism, 86, 88–90
societal disconnectedness, 48
Risk markers, 56
S
Saudi Arabia, 88
Search strategy, 35
Security agencies, 87, 105
Security checks, 287–289
Security offences, 90
Security offenders, 86, 91, 98
Security personnel, 302
Security prisoners, 90, 92, 98
Security threats, 150
Segregation, 206
Self-control theory, 70
Self-identified Muslims survey in UK, 116
Self-reinforcing mechanisms, 7
SEM models, 46
Sentence length, 89
Sex offenders, 86
Shields of resilience, 57
Sicilian Mafia, 251, 254
“Significance loss”, 152
Silence/omertà, 196
Smuggling methods, 278, 284, 286, 303, 304
Snowball method, 59
Social bonds, 12, 293, 295, 301
Social information processing, 70
Social learning theory, 69, 70
Social mobility, 7, 11
vs. income inequality, 237
inequality, 237
OC, 205, 207, 237
Social opportunity structure, 311
Social/psychological/economic factors, OCGs
age, 190
criminal background and skills, 195
economic conditions, 192, 193
educational background, 192
Index
employment, 192
ethnicity, 191
gender, 191
group identity, 194
silence/omertà, 196
social ties, 193, 194
Social-psychological theories, 32
Social relations, 294
Social snowball effect, 175
Social snowball method, 298
Social ties, 175, 193, 194, 279
Socio-demographic characteristics, 112
Socio-demographic factors, 43
Socio-economic determinants
anonymized database, 155
BRP, 155
datasets, Statistic Netherlands, 156
demographic descriptives, 156
descriptive statistics, 155
Dutch prosecution office, 155
education levels, 157
identifiable data, 155
logistic regression analyses, 155
logistic regression models, 158, 159
non-immigrant background, 158
origin, SES and criminality, 157
terrorism suspects and factor, 158
variables, 156
Socio-economic factors, 13, 15
crime-terror nexus, 152
disadvantage, 152
Dutch terrorism, 152
group categorization, 153
group violence, 153
Jihadi travellers, 152
and marginalization, 152
perceived threat, 153
personal deprivation and strains, 153
social milieu, 152
unemployment, 152
Socio-economic mobility
cross-sectional regressions, 217, 232
inequality, 215, 216
OC, 227
Socio-economic status (SES), 43, 48, 153
Socio-economic variables, 162, 163
Socio-political context, 149–151
Socio-psychological factors, 6
Strain theory, 69
Street stops vs. domestic raids, 116
Structural equation modelling (SEM), 41
Support for terrorism, 31
Survey on Household Income and Wealth
(SHIW), 208
Index
Suspect community, 106
Systematic reviews
ability, 32
democratic vs. non-democratic
countries, 34
development, 28
effectiveness of counter-terrorism
strategies, 28
extensive and rigorous search, 198
protective factors, 66
qualitative methods, 196
quantitative methods, 196
quantitative results, 58
radicalization, 26, 27
research designs, 34, 35
research questions, 177
search process, 59, 60
search strategy, 179, 180
search terms, 59
social and economic factors, 196
standards, 31
Systematic screening, 179
T
Terrorism
attitudinal and behavioral aspects, 148
criminological perspective, 97
definition, 5, 55
explosion, 25
goals, 2
government, 1
impacts, 2
instruments, 57
intentional act/threat of violence, 55
literature (see Literature)
macro-level predictors, 34
multi-disciplinary attention, 27 (see also
Organized crime (OC))
police legitimacy, 14
prior criminality, 14
and procedural injustice, 107–110
radicalization, 5, 6
re-categorized/re-analyzed, 27
recidivism, 9
recruitment, 5, 6
researchers, 27
social organization, 1
socio-demographic factors, 4
systematic reviews, 8
types, 71
unemployment, 6
in US, 122
and violent groups, 69
347
Terrorism Act 2000, 110
Terrorism Act 2006, 110
Terrorism offences, 87, 89, 98
Terrorism offenders, 85, 87, 89, 90, 97, 98,
100, 101
Terrorism offending, 88–90
Terrorism policing, 109
Terrorism-related crime, 122
Terrorism-related offences, 98
Terrorism-related research, 121
Terrorism Situation and Trend Report
(TE-STAT), 88
Terrorism units, 160
Terrorist Crime Act, 149
Terrorist ideology, 123, 127
Terrorist offenses, 148, 149, 155–158, 160,
162, 166, 167
Terrorist organizations, 90
Terrorist suspects, 148, 149, 151, 152,
155–162, 166–168
Terrorist units, 149, 160–165, 168
Theory of Planned Behavior model, 32
Threat management system, 153
Threat model of terrorism
categorization, 149
discrimination/segregation, 154
distal and proximal sources, 154
personal zone, 154
psychological factors (see Psychological
factors)
socio-economic determinants,
155–160
socio-economic factors (see Socio-­
economic factors)
socio-political context, 149–151
suicide terrorism, 155
Traditional criminogenic, 9
Transit crime
Amsterdam Airport Schiphol, 281
deliberate recruitment, 279
description, 278
drug trafficking (see Drug trafficking)
EMCDDA, 280
leisure activities and sidelines, 280
life events, 279, 280
Netherlands, 279–281, 310
offender convergence
settings, 280
Port of Rotterdam, 282
social bonding, 279
social ties, 279
transport drugs, 280
work ties, 279
Trustworthy motives, 108
Index
348
T-tests
perceived risk, terrorism recruitment,
114, 115
Two-Pyramid Model’s, 32
U
Unbiased decision makers, 108
Unemployment, 152, 162
Unintended negative outcomes, 105
United Nations Office on Drugs and Crime
(UNODC), 2
U.S. extremists
age of crime onset, 134
pre-radicalization crime rates,
129–131
pre-radicalization criminal
activity, 131–133
V
VERA-2, 57
Violence-justifying beliefs, 135
Violence-permissive attitudes, 149
Violent action, 149
Violent activity, 29
Violent and non-violent terrorism-related
offences, 86
Violent behaviour, 61, 99
Violent crimes, 122, 123, 125, 126, 130,
132–134, 136, 140
Violent extremism, 28, 32, 121–124, 126, 129,
133, 134, 136–142
Violent offenders, 99
W
War on terrorism, 30